acthub.beta

OFCOM’s guidance about children’s risk assessments

universal formatas at 12 May 2025124 references

OFCOM’s guidance about children’s risk assessments

Children’s Risk Assessment Guidance and Children’s Risk Profiles

Published 24 April 2025

Contents

Part 1: Duties and carrying out a children’s risk assessment

1. Introduction ........................................................................................................................ 4
2. Children’s risk assessment duties....................................................................................... 6

Part 2: How to carry out a children's risk assessment

1. Overview of the four-step risk assessment process ........................................................ 18
2. Detailed explanation of the four steps............................................................................. 21

Part 3: Supporting information

1. Children’s Risk Profiles ..................................................................................................... 38
2. Non-designated content .................................................................................................. 56
3. Evidence inputs ................................................................................................................ 58
4. Risk Level Table for content harmful to children ............................................................. 68
5. Making a significant change to your service .................................................................... 72

Annex

1. Appendix A: Examples of how to use the Risk Level Table .............................................. 76
6. Appendix B: Comparison of children’s and illegal content risk assessment duties ......... 64

Part 1: Duties and carrying out a children’s risk assessment

Section 1 Introduction

1.1 This guidance aims to help providers of services likely to be accessed by children to comply with the children’s risk assessment duties, as set out in the Online Safety Act 2023 (the Act).1 The purpose of the children’s risk assessment is to improve your understanding of the risk of harm to children on your service, and what safety measures you need to put in place to protect them.
Box 1: Assessing the risk of harm to children To meet the requirements of the children’s risk assessment duties set out in the Act, you should assess the risk of harm to children presented by content that is harmful to children. Content harmful to children includes primary priority content, priority content, and non-designated content that is harmful to children. You must give separate consideration to children in different age groups and also assess how the design and use of your service affects the level of risk of harm to children.
1.2 All providers of Part 3 services must complete a children’s access assessment; this will help you to work out if your service, or part of your service, is likely to be accessed by children. Please see Ofcom’s Children’s Access Assessments Guidance. All providers of services likely to be accessed by children must complete a children’s risk assessment.
1.3 The children’s risk assessment is separate, and additional, to the illegal content risk assessment required for all Part 3 services.2
1.4 This guidance recommends the same four-step methodology as in Ofcom’s Illegal Content Risk Assessment Guidance for Service Providers, adapted as appropriate to reflect the children’s risk assessment duties. Following the four-step process in this guidance will help you to comply with the children’s risk assessment duties, and the associated safety duties and record-keeping and review duties.
1.5 The guidance consists of three parts and two appendices: a) Part 1 includes a summary of the children’s risk assessment duties. b) Part 2 includes an overview of the four-step risk assessment methodology, and detailed information on completing the children’s risk assessment, including:
i The four-step methodology: We recommend that all service providers follow these steps to carry out risk assessments. Following this can also help you to meet other duties under the Act.
ii What evidence to use: Risk assessments require evidence to ensure they are suitable and sufficient. We expect all in-scope service providers to consult ‘core’ evidence inputs, and some providers to use additional ‘enhanced’ evidence inputs where this is available or appropriate.
iii When to review or carry out a new risk assessment: Guidance on how to keep a risk assessment up to date and the circumstances under which you need to carry out a new assessment.
1 Section 11 for user-to-user services and section 28 for search services in the Act. 2 Section 7(4) of the Act. For illegal content risk assessment duties, see section 9 for user-to-user services and section 26 for search services in the Act.
c) Part 3 includes supporting information for completing each step in the four-step methodology. d) Appendix A provides examples of how to use the Risk Level Table. e) Appendix B contains a comparison of the children’s and illegal content risk assessment duties.
1.6 We also provide supporting resources and tools on the Ofcom website to help you to understand how to meet your obligations.

Section 2 Children’s risk assessment duties

2.1 This sub-section of the guidance explains the children’s risk assessment duties.

What are the children’s risk assessment duties?

2.2 If you provide a user-to-user or search service that is likely to be accessed by children, you must carry out a children’s risk assessment.3 Broadly speaking, this legal obligation requires you to assess the risk of harm your service poses to children presented by content harmful to children4 set out in the Act, giving separate consideration to children in different age groups, and assessing how the design and use of your service affects the level of risk of harm to children.5
2.3 The Act sets out the specific elements that a children’s risk assessment needs to include.6 You need to take into account your service’s characteristics, such as its user base and functionalities.7 You must take account of Ofcom’s Children’s Risk Profiles, which identify the features or functionalities that could lead to a risk of harm to children on your service if you have them.8 These are separate and distinct from the Risk Profiles published alongside our Illegal Content Risk Assessment Guidance.
2.4 Your children’s risk assessment must be suitable and sufficient,9 which we explain in more detail in the sub-section ‘What is a ‘suitable and sufficient’ children’s risk assessment?’
2.5 If you have more than one service that is likely to be accessed by children, you need to complete a separate children’s risk assessment for each service. It is your responsibility to determine the distinct services it has which are likely to be accessed by children and the boundaries between these services.
2.6 Once you have completed your first children’s risk assessment, there are several duties in the Act related to reviewing, updating or completing new children’s risk assessments: • You must take appropriate steps to keep your children’s risk assessment up to date, including when Ofcom makes a significant change to a Children’s Risk Profile that relates to your service;10
3 Sections 7(4), 11(1), 24(4) and 28(1) of the Act. 4 The legal definition of content harmful to children is included in section 60 of the Act.
5 Sections 11 and 24 of the Act. 6 Children’s risk assessment duties are set out in Section 11 of the Act (for user-to-user service providers), and section 28 of the Act (for search service providers). 7 Section 11(6) and 28(5) of the Act. 8 Section 11(6) and 28(5) of the Act. 9 Section 11(2) and 28(2) of the Act. 10 Sections 11(3) and 28(3) of the Act.
• You must carry out a further children’s risk assessment before making any significant change to any aspect of your service’s design or operation, relating to that proposed change;11 • You need to keep a record of each children’s risk assessment you carry out.12 We have published separate Record-Keeping and Review Guidance to help you with this.
2.7 To help you meet these duties, our guidance sets out a four-step risk assessment process. Following this process will help you comply with the children’s risk assessment duties, and the linked safety duties and record-keeping duties. We recommend that you also consult the detailed research and analysis of content harmful to children in Ofcom’s Children’s Register of Risks (Children’s Register).13

What is a ‘suitable and sufficient’ children’s risk assessment?

2.8 Risk assessments must meet the requirements of the children’s risk assessment duties, including the ‘suitable and sufficient’ standard.
2.9 To be suitable and sufficient, your children’s risk assessment must include all the elements of a children’s risk assessment specified in the Act.14 This means it should be specific to your service and reflect the risks accurately. This will lead you to an adequate understanding of risks to children on your service to implement appropriate safety measures to mitigate those risks and protect children on your service.
2.10 Your assessment must: a) Assess the risk of children encountering each kind of primary priority content, each kind of priority content, and non-designated content; b) Take into account Ofcom’s Children’s Risk Profiles;15 c) Consider the characteristics of your service: its user base (e.g., user numbers, different age groups of children, languages, groups at risk, and groups increasing risk), functionalities (including those that enable adults to search for or contact other users, including children), algorithmic systems (and how easily, quickly and widely they disseminate content) and the business model; d) Consider any other relevant aspects of your service’s design and operation, including any existing controls to mitigate harm such as governance, use of proactive technology, measures to promote users’ media literacy and safe use of your service, and other systems and processes which could affect the level of risk to children; and e) Consider how your service is used – for example, both the intended and unintended ways that children may use the service, and functionalities and features that affect how much children use the service.
11 Sections 11(4) and 28(4) of the Act. We explain what a significant change may involve in the sub-section ‘Making a significant change to your service’.
12 Sections 23(2) and 34(2) of the Act. 13 The Children’s Register is our assessment of the causes and impacts of harms to children online based on the evidence that we have gathered over the past four years. 14 Section 11(6) for user-to-user services and section 28(5) for search services. 15 These are set out in Part 3, sub-section 1 and include a list of risk factors (such as features and functionalities) with an explanation of how they could increase the risk of particular content harmful to children covered by the Act.
2.11 You should make a judgement about the risk level for each of the four kinds of primary priority content, each of the eight kinds of priority content, and for any kind(s) of non-designated content that you identify for assessment on your service (see Table 1). Our detailed guidance helps you to do this by assessing the likelihood of children encountering content harmful to children by means of your service, and the impact of this on children. It also explains how you should identify the kind(s) of non-designated content that you need to assess on your service.
2.12 Your assessment should be based on relevant information and evidence to accurately reflect the risks of harm to children so that you can keep a record of your analysis and conclusions. Services which have complex risk factors, characteristics, design features or different user types may require more types of evidence to inform their judgements of risk.
2.13 Our guidance on evidence inputs sets out: a) The core evidence inputs that all service providers should consider, including information on your service’s characteristics and risk factors (from our Children’s Risk Profiles), user complaints including user reports, relevant user data, and any other relevant information you already hold, such as relevant findings from your children’s access assessment, and from your illegal content risk assessment. Such information should be readily available to all service providers, and failing to consider it may mean that the children’s risk assessment is not suitable and sufficient. b) Service providers should consider additional enhanced evidence inputs where the core inputs are insufficient to make accurate assessments.16 Examples of enhanced inputs include: results of product testing; insights from content moderation systems; consultation with technical or independent experts, or with users and representative groups; research into children’s behaviours and needs; or external audit and risk assurance processes. See the sub-section ‘Evidence inputs’ in Part 3, for guidance on the kinds of evidence that should be collected and assessed.

Record-keeping duties

2.14 You need to keep a record of each children’s risk assessment you carry out, including details of how the assessment was carried out and its findings.
2.15 We have published separate Record-Keeping and Review Guidance to help providers understand what is expected of them and what to record, and we provide further information in Part 2 of this section, in the sub-section ‘Step 2: Assess the risk of harm to children’.

Additional duties for categorised service providers

2.16 Ofcom will publish a register of categorised services.17 These services will have additional duties relating to their children's risk assessments. Category 1 and Category 2A service providers must: a) Publish a summary of their most recent children’s risk assessment in their terms of service (Category 1) or in a publicly available statement (Category 2A). The summary
16 Providers of large services, or those who identify several specific risk factors for a kind of content harmful to children should consider using enhanced inputs to achieve a suitable and sufficient assessment. 17 Section 95(10) of the Act.
must include the findings of the most recent children’s risk assessment of a service (including as to levels of risk and as to nature, and severity, of potential harm to children);18 and
b) Provide Ofcom with a copy of their children’s risk assessment record as soon as reasonably practicable.19

What happens if you do not carry out a suitable and sufficient risk assessment?

2.17 If we suspect that you have failed to carry out a suitable and sufficient children’s risk assessment, then we are able to take enforcement action.20 Any decision we take regarding enforcement action would be made in line with Ofcom’s Online Safety Enforcement Guidance.
2.18 If we decide to open an investigation and find that your service has failed to comply with its duties, we may impose a penalty of up to 10% of qualifying worldwide revenue or £18 million (whichever is the greater) and require remedial action to be taken.21

When do you need to complete the first children’s risk assessment?

2.19 All providers of Part 3 services must carry out children’s access assessments. Your children’s access assessment will help you to work out if your service, or part of your service, is likely to be accessed by children. Please see our Children’s Access Assessments Guidance.
2.20 If your service is treated as likely to be accessed by children on or before 24 April 2025, you will have until 24 July 2025 to complete your first children’s risk assessment.
2.21 If you start operating a new Part 3 service or carry out a new children’s access assessment,22 and conclude the service is likely to be accessed by children, you must complete your children’s risk assessment within three months after concluding the children’s access assessment.
2.22 There are also duties about reviewing, updating or carrying out a new children’s risk assessment. In summary: • You must take appropriate steps to keep a children’s risk assessment up to date, including if Ofcom makes a significant change to a Children’s Risk Profile that relates to your service; and • You must carry out a further children’s risk assessment before making a significant change to any aspect of your service’s design or operation.
18 Section 12(14) of the Act (user-to-user service providers) and Section 29(9) of the Act (search service providers). 19 Section 23(10) of the Act (user-to-user service providers) and Section 34(9) of the Act (search service providers). 20 Section 131 of the Act. 21 Section 143 of the Act and schedule 13 to the Act. 22 See Section 5 of the Children’s Access Assessments Guidance
2.23 See the sub-section ‘Making a significant change to your service’ in Part 3 of this guidance document for further details on what constitutes a significant change.

What content harmful to children do you need to assess?

2.24 Carrying out a children’s risk assessment requires you to assess the risk of harm to children on your service presented by content that is harmful to children. Content harmful to children includes primary priority content that is harmful to children, priority content that is harmful to children and non-designated content that is harmful to children.23
2.25 You should assign a risk level to each of the four kinds of primary priority content, each of the eight kinds of priority content, and for any kind(s) of non-designated content you identify for assessment on your service, as presented in Table 1.

Table 1: List of content harmful to children to assess

Primary priority content
• Pornographic content. • Content which encourages, promotes or provides instructions for suicide. • Content which encourages, promotes or provides instructions for an act of deliberate self-injury. • Content which encourages, promotes, or provides instructions for an eating disorder or behaviours associated with an eating disorder.
Priority content
• Content which is abusive and which targets any of the following characteristics – (a) race, (b) religion, (c) sex, (d) sexual orientation, (e) disability or (f) gender reassignment. • Content which incites hatred against people – (a) of a particular race, religion, sex or sexual orientation, (b) who have a disability, or (c) who have the characteristic of gender reassignment. • Content which encourages, promotes or provides instructions for an act of serious violence against a person. • Bullying content. • Content which – (a) depicts real or realistic serious violence against a person; (b) depicts the real or realistic serious injury of a person in graphic detail. • Content which – (a) depicts real or realistic serious violence against an animal; (b) depicts the real or realistic serious injury of an animal in graphic detail; (c) realistically depicts serious violence against a fictional creature or the serious injury of a fictional creature in graphic detail.
• Content which encourages, promotes, or provides instructions for a challenge or stunt highly likely to result in serious injury to the person who does it or to someone else. • Content which encourages a person to ingest, inject, inhale or in any other way self-administer – (a) a physically harmful substance; (b) a substance in such a quantity as to be physically harmful.
Non-designated content
Content harmful to children that is not primary priority content or priority content, and is of a kind which presents material risk of significant harm to an appreciable number of children in the UK. There are some exceptions to what could be considered non-designated content.24 Ofcom has identified two kinds of non-designated content that meet the above criteria:25 • Content that shames or otherwise stigmatises body types or physical features (‘body stigma content’). • Content that promotes depression, hopelessness and despair (‘depression content’). Services may identify other kinds of content that they consider meet the definition of non-designated content that is harmful to children, in which case they should include those kinds of content in their children’s risk assessment and assign a risk level to them.26
23 Sections 60-62 of the Act.
2.26 You can consult Ofcom’s Children’s Register for research and analysis on each kind of content harmful to children to support your assessment. Ofcom’s Guidance on Content Harmful to Children can also help providers to identify content harmful to children.
2.27 Some sections in the Children’s Register have grouped different content harmful to children, such as suicide and self-harm.27 The Children’s Risk Profiles have grouped these in line with the Children’s Register. Regardless of how evidence is presented in the Children’s Register and Children’s Risk Profiles, service providers have a duty to assess the risk of children encountering each kind of primary priority content, each kind of priority content, and non-designated content.

What to assess about each kind of content harmful to children

2.28 Service providers need to assess the risk of harm to children presented by content harmful to children on their service.
24 Content is not to be regarded as non-designated content where the risk of harm flows from the content’s potential financial impact; the safety or quality of goods featured in the content; or the way in which a service featured in the content may be performed. 25 See the Children’s Register for our evidence on these types of non-designated content. 26 If you have identified those kinds of non-designated content as present on your service, you will have an additional duty to notify Ofcom of the presence and incidence of that content: see Step 4. 27 See the Children’s Register.

Harm

2.29 ‘Harm’ is defined in the Act as physical or psychological harm.28 This includes circumstances where, as a result of the content, individuals act in a way that results in harm to themselves or that increases the likelihood of harm to themselves, and circumstances where, as a result of the content, individuals do or say something to another individual that results in harm to that other individual or that increases the likelihood of such harm.
2.30 As evidenced in the Children’s Register, harm also includes cumulative harm that arises where:29 • Content, or content of a particular kind, is repeatedly encountered by an individual (including, but not limited to, where content, or a kind of content, is sent to an individual by one user or by different users or encountered as a result of algorithms used by, or functionalities of, a service); or • Content of a particular kind is encountered by an individual in combination with content of a different kind (including, but not limited to, where a kind of content is sent to an individual by one user or by different users or encountered as a result of algorithms used by, or functionalities of, a service).
2.31 Additionally, harm can include circumstances where:30 • Individuals act in a way that results in harm to themselves, or increase the likelihood of harm to themselves; or • Individuals do or say something to another individual that results in harm to that other individual, or increases the likelihood of such harm (including, but not limited to, where individuals act in such a way as a result of content that is related to that other individual’s characteristics or membership of a group).

Assessing the risk of harm

2.32 To meet the requirements of the children’s risk assessment set out in the Act, you should assess the risk of harm to children arising from each kind of primary priority content, each kind of priority content, and any kind(s) of non-designated content you have identified for assessment, giving separate consideration to children in different age groups, and assessing how the design and use of the service affects the level of risk of harm to children.
2.33 In the assessment, you must consider all the requirements in section 11 (for user-to-user services) or section 28 (for search services) of the Act, and should use evidence about your service to make your assessment.
2.34 You should then assign a risk level to each of the four kinds of primary priority content, each of the eight kinds of priority content, and to any kind(s) of non-designated content you identify for assessment on your service: negligible, low, medium or high. Determining a risk level requires evaluating the likelihood and impact (such as the nature and severity) of children encountering each kind of primary priority content, each kind of priority content, and any kind(s) of non-designated content you identify for assessment on your service; this is explained in Step 2 of the four-step methodology in this guidance.
28 Section 234 of the Act. 29 Children’s Register of Risks. 30 Section 234 of the Act.
2.35 If you provide a search service, you must consider the risk of children who are users of the service encountering ‘search content’ that is harmful to children.31 Search content is content that may be encountered in or via search results of search services, for example, content presented to a child user of the service by operation of the search engine in response to a search request made by a child.

Using Children’s Risk Profiles and the Risk Level Table in your children’s risk assessment

2.36 To help you carry out your children’s risk assessment, you must take account of the Children’s Risk Profile relevant to your service.32 Ofcom has a duty to prepare and publish Children’s Risk Profiles,33 setting out the risk factors (features and functionalities) that are most strongly linked to content harmful to children. These will help you to identify and record the relevant risk factors for your service. There are two sets of Children’s Risk Profiles – one for user-to-user services and one for search services. These are set out in Part 3, sub-section ‘Children’s Risk Profiles’.
2.37 The guidance intends to help you to assess whether your service has a negligible, low, medium or high level of risk for each kind of primary priority content, each kind of priority content, and for any kind(s) of non-designated content you identify on your service. To determine the level of risk for each particular kind of content, you must consider the Children’s Risk Profiles, and make an assessment of the level of risk to children based on relevant evidence inputs, taking account of the impact of any existing controls on your service.
2.38 Service providers can use their own risk assessment methodology. However, we have produced a Risk Level Table to support you in reaching an assessment of the level of risk for each particular kind of content harmful to children, taking into account the Children’s Risk Profiles and other relevant factors or evidence.
2.39 If you use your own risk assessment methodology, you are still expected to justify separate conclusions of high, medium, low or negligible risk for each kind of primary priority content, each kind of priority content, and for any kind(s) of non-designated content you have identified for assessment on your service. This is to ensure that you keep a record of all assessments of levels of risk that relate to any measures in the Protection of Children Codes.

Keeping your children’s risk assessment up to date

2.40 After you have completed a first risk assessment, you must take appropriate steps to keep your children’s risk assessment up to date, including by responding to key triggers.
2.41 The Act includes the following duties about reviewing or carrying out a new risk assessment, to be considered as key triggers: a) A duty to take appropriate steps to keep a children’s risk assessment up to date;
31 Search content is defined in section 57 of the Act. 32 Sections 11(6) and 28(5) of the Act. 33 Section 98(5) and (7) of the Act.
b) A duty to update your children’s risk assessment if Ofcom makes any significant change to a Children’s Risk Profile that relates to your service; and c) Before making any significant change to any aspect of your service’s design or operation, a duty to carry out a further suitable and sufficient children’s risk assessment relating to the impacts of that proposed change.
2.42 Depending on circumstances, it may be appropriate to fully review your children’s risk assessment (e.g., as a matter of course at least every year), or target the review to specific aspects of your service (e.g., due to the introduction of a new feature or functionality outside a regular annual review cycle).
2.43 We recognise that some service providers may want to consider the timing of their children’s risk assessments in the context of their obligations under other legal regimes (such as the Digital Services Act (DSA) (EU)), or privacy and data protection impact assessments). This is a matter for individual service providers, provided that the approach they take meets their obligations under the Act.

Review and update at least every 12 months

2.44 You have a duty to take steps to keep your assessment up to date.34 We recommend that children’s risk assessments are reviewed at least every 12 months.
2.45 It is likely that your risk assessment will become out of date after a certain amount of time has passed, even if you have not made any significant changes to your service. Incremental changes to your service, trends in child users’ behaviours, and technological changes will alter the evidence base underpinning your assessment, which would require you to review your assessment.
2.46 Reviewing your assessment can be done by considering your most recent children’s risk assessment alongside any new evidence that you have collected during the operation of your service, or new developments in the external environment and the risks online which your assessment needs to account for. If you think the new evidence may impact your assessment of risk to children, then you should review each step of the process with the updated evidence.
2.47 You should decide your own policy for reviewing the children’s risk assessment and updating it. You should be able to explain your approach and what appropriate steps you are taking to meet this duty. A written policy is a valuable tool to help you to demonstrate compliance. You will also need to keep a written record of any updates to the children's risk assessment in line with our Record-Keeping and Review Guidance.
2.48 Your written policy on keeping a children’s risk assessment up to date should include: • A timeframe for regular review: Ofcom recommends that children’s risk assessments are reviewed at least every 12 months. This aligns with other common annual governance, reporting and compliance cycles, and with other international online safety frameworks (including the DSA (EU)); and • A responsible person overseeing children’s risk assessment processes.
34 Sections 11(3) and (4), and 28(3) and (4) of the Act.

Review and update if Ofcom makes a change to Children’s Risk Profiles

2.49 The Act requires Ofcom to review and revise our Children’s Register and Children’s Risk Profiles to keep them up to date.35
2.50 If Ofcom makes a significant change to a Children’s Risk Profile which is relevant to your service, you must review and update your children’s risk assessment.36
2.51 This can be done by considering your most recent children’s risk assessment alongside Ofcom’s changes to the Children’s Risk Profiles, to understand if any aspect of your assessment needs to be updated. For example, new risk factors relevant to your service could have been added (such as new functionalities), or new links between risk factors and content harmful to children could have been identified (such as a new functionality which increases the risk of content which encourages, promotes, or provides instructions for suicide). You can sign up for updates from Ofcom to be made aware of changes to Children’s Risk Profiles on Ofcom’s website.
2.52 If these changes are relevant to your service, you should consider if your assessment of the risk of content harmful to children needs to change.

Relevance of the illegal content risk assessment

2.53 The Act sets out that all providers of Part 3 services must conduct an illegal content risk assessment. Therefore, you must assess the risks on your service associated with priority illegal content and other illegal content, including relevant non-priority illegal content.37 You may decide to carry out your illegal content and children’s risk assessments concurrently. If you do this, you should be mindful of the following: • Service providers must carry out risk assessments for both duties, and hold separate records for each risk assessment. • Service providers will identify different Codes measures to reduce the risk of each kind of illegal content or content harmful to children, as a result of the respective risk assessments.
2.54 When considering evidence and making assessments, you should note that some categories of content harmful to children may be related to or overlap with some categories of illegal content.38 For example, certain content that promotes, encourages or provides instructions for suicide may, in addition to being primary priority content, also amount to priority illegal content (encouraging or assisting suicide) for the purposes of the illegal content risk assessment.
2.55 Similarly, some characteristics of the service that affect the level of risk of illegal harm (such as user base, functionalities, and the ways in which a service is used) are also likely to be relevant for your children’s risk assessment.
35 Section 98(8) of the Act. 36 Sections 11(3) and 28(3) of the Act. 37 Section 59(2) of the Act. 38 If a service provider believes content could count as illegal content, they should apply the illegal content duties and take note of the children’s safety duties.
2.56 Should you conduct risk assessments concurrently, you still need to ensure that the illegal content risk assessment and the children’s risk assessment are distinct and clearly identifiable.
2.57 Finally, the findings of your illegal content risk assessment may inform your children’s risk assessment – for example, evidence on hateful content that may be illegal may also support your assessment of content that incites hatred.

Part 2: How to carry out a children’s risk assessment

Section 3 Overview of the four-step risk assessment process

Step 1: Understand content harmful to children that needs to be assessed

Step 1 will help you to understand the content harmful to children to assess and will prepare you to make accurate judgements about your risk to children.

Sequence of activities and outcomes

• Identify the content harmful to children that needs to be separately assessed, including each kind of primary priority content, each kind of priority content, and any kind(s) of non-designated content that may be relevant to your service. • Consult Ofcom’s Children’s Risk Profiles and identify the key risk factors relevant to your service for each kind of primary priority content, priority content, and the kind(s) of non-designated content identified by Ofcom.

Essential records

• Confirmation that you have consulted Ofcom’s Children’s Risk Profiles and recorded any risk factors relevant to your service. • A record of the kind(s) of non-designated content that will be assessed, and how you have considered the risk of children encountering non-designated content by means of the service.

Step 2: Assess the risk of harm to children

Step 2 will help you use evidence to assess and assign a risk level to the risk of harm to children presented by each of the four kinds of primary priority content, each of the eight kinds of priority content, and any kind(s) of non-designated content you have identified for assessment. When assigning these risk levels, you must also consider the impact of your service’s design and use on the risk of harm.

Sequence of activities and outcomes

• Separately assess the likelihood and impact of children encountering the content harmful to children you have identified at Step 1 for assessment on your service, using all relevant evidence. • As part of your assessment of likelihood and impact: o Consider the different ways in which the service is used, including ways which are unintended. Identify whether there are any additional characteristics or functionalities of the service’s design or operation, not in the Children’s Risk Profiles, which could increase the risk to children. This includes functionalities that present higher levels of risk such as recommender systems, or those that enable adults to
search and/or contact children, predictive search functionalities, and features and functionalities which affect how much children use the service.
o Consider the effectiveness of any existing control measures which could impact the level of risk of harm to children.
• Consult the Risk Level Table to assign a risk level for each of the four kinds of primary priority content, each of the eight kinds of priority content, and any kind(s) of non-designated content you have identified for assessment. This risk level should reflect risk as it exists on the service at the time of assessment, having had regard to the efficacy of any existing control measures you have in place. • Conclude the assessment of all the risks relating to content harmful to children, including the design and use of the service, to mitigate in Step 3.

Essential records

• Where applicable, a list of any additional characteristics (including user base, business models, functionalities, governance, and systems and processes) considered alongside the risk factors identified in Ofcom’s Children’s Risk Profiles. This should include how you have identified and assessed those functionalities that present higher levels of risk such as recommender systems, functionalities which enable adults to search for and/or contact children and predictive search functionalities, as well as features and functionalities which affect how much children use the service. • If you have considered the role of any existing controls already in operation on your service at the time of this risk assessment, you should record what these controls are, what risks they are intended to mitigate and how they do this, and how the consideration of the existing controls has impacted the risk level you have assigned to a kind of content harmful to children. • A list of the evidence and summary of the reasoning that has informed the assessment of likelihood and impact of each kind primary priority content, each kind of priority content, and any kind(s) non-designated content that you have identified for assessment. This should include relevant evidence about the design and use of the service, and how different age groups of children have been considered. • The level of risk assigned to each of the four kinds of primary priority content, each of the eight kinds of priority content, and for any kind(s) of non-designated content that you have identified for assessment on the service, and an evidence-based explanation of the decision. This level should reflect risk as it exists on the service at the time of assessment.

Step 3: Decide measures, implement and record

Step 3 will help you identify any relevant measures to implement to address risk to children, record any measures you have taken, and make a record of your assessment.

Sequence of activities and outcomes

• Consult Ofcom’s Protection of Children Codes for user-to-user services and search services , check which measures are recommended for your service, and decide whether to implement applicable measures to reduce the risk of harm to child users, or use alternative measures.
• Identify any additional measures that may be appropriate for your service. • Implement all relevant measures. • Record the outcomes of the children’s risk assessment.

Essential records

• A complete record of the findings of the children’s risk assessment. • All measures from Ofcom’s Protection of Children Codes that have been, or are planned to be, implemented. • Any applicable measures from Ofcom’s Protection of Children Codes that are not planned to be implemented, and the alternative measures that have been or are planned to be implemented instead, with information that demonstrates how these alternative measures meet the relevant duties.

Step 4: Report, review and update

Step 4 will help you to understand how to keep your children’s risk assessment up to date and put in place appropriate steps to review your assessment.

Sequence of activities and outcomes

• Report on the children’s risk assessment and measures through appropriate governance and accountability channels. • Providers of user-to-user services to notify Ofcom of the kinds and incidence of any non-designated content you have identified as present on your service through your children’s risk assessment. • Providers of Category 1 and 2A services to supply Ofcom with a copy of their children’s risk assessment record. • Providers of Category 1/2A services to summarise the findings of their most recent children’s risk assessment in their terms of service/a publicly available statement (as applicable). • Monitor the effectiveness of safety measures at reducing the risk of harm to users. • Monitor developing risks and the level of risk exposure after appropriate measures are implemented (also known as residual risk). • Review and/or update the children’s risk assessment when appropriate, including before making any significant change to any aspect of the service’s design or operation.

Essential records

• A written record of the annual review cycle for the children’s risk assessment, and the named person responsible who has been appointed for this process. • Confirmation that the findings of the children’s risk assessment have been reported, and recorded, through appropriate governance and accountability channels.

Section 4 Detailed explanation of the four steps

This sub-section explains in detail the activities and outcomes for each step of your children’s risk assessment.

Step 1: Understand content harmful to children that needs to be assessed

4.1 The objectives of this step are for service providers to identify and develop an understanding of content harmful to children that they need to assess, and to consult Ofcom’s Children’s Risk Profiles.

Sequence of activities and outcomes for Step 1

Identify content harmful to children you need to assess

4.2 The Act distinguishes between primary priority content, priority content and non-designated content that is harmful to children.39
4.3 All service providers are required to assess the risk of children encountering each kind of primary priority content, each kind of priority content, and non-designated content by means of their service.40
4.4 See Table 2 for the list of content harmful to children. As part of the first step of your children’s risk assessment, we recommend that you familiarise yourself with this list.

Table 2: List of content harmful to children to assess

Primary priority content
• Pornographic content. • Content which encourages, promotes or provides instructions for suicide. • Content which encourages, promotes or provides instructions for an act of deliberate self-injury. • Content which encourages, promotes, or provides instructions for an eating disorder or behaviours associated with an eating disorder.
Priority content
• Content which is abusive and which targets any of the following characteristics – (a) race, (b) religion, (c) sex, (d) sexual orientation, (e) disability or (f) gender reassignment.
• Content which incites hatred against people – (a) of a particular race, religion, sex or sexual orientation, (b) who have a disability, or (c) who have the characteristic of gender reassignment. • Content which encourages, promotes or provides instructions for an act of serious violence against a person. • Bullying content. • Content which – (a) depicts real or realistic serious violence against a person; (b) depicts the real or realistic serious injury of a person in graphic detail. • Content which – (a) depicts real or realistic serious violence against an animal; (b) depicts the real or realistic serious injury of an animal in graphic detail; (c) realistically depicts serious violence against a fictional creature or the serious injury of a fictional creature in graphic detail. • Content which encourages, promotes, or provides instructions for a challenge or stunt highly likely to result in serious injury to the person who does it or to someone else. • Content which encourages a person to ingest, inject inhale or in any other way self-administer – (a) a physically harmful substance; (b) a substance in such a quantity as to be physically harmful.
Non-designated content
Content harmful to children that is not primary priority content or priority content, and is of a kind which presents material risk of significant harm to an appreciable number of children in the UK. There are some exceptions to what could be considered non-designated content.41 Ofcom has identified two kinds of non-designated content that meet the above criteria:42 • Content that shames or otherwise stigmatises body types or physical features (‘body stigma content’). • Content that promotes depression, hopelessness and despair (‘depression content’). Services may identify other kinds of content that they consider meet the definition of non-designated content that is harmful to children, in which case they should include those kinds of content in their children’s risk assessment and assign a risk level to them.43
39 These types of content are set out in sections 60-62 of the Act. 40 You can find more detail on identifying non-designated content in the sub-section ‘Non-designated content’ in Part 3 of this guidance document.

What to assess about each kind of content harmful to children

4.5 Your children’s risk assessment must separately assess the risk of children encountering each kind of primary priority content, each kind of priority content, and non-designated content by means of your service.
41 Content is not to be regarded as non-designated content where the risk of harm flows from the content’s potential financial impact; the safety or quality of goods featured in the content, or the way in which a service featured in the content may be performed. 42 Please see the Children’s Register for our evidence on these types of non-designated content. 43 If you have identified those kinds of non-designated content as present on your service, you will have an additional duty to notify Ofcom of the presence and incidence of that content: see Step 4.
4.6 The Children’s Register and the Guidance on Content Harmful to Children include useful information that should help you understand more about the risks to children online.44
4.7 The primary objective is for you to consider how your service may be used in a way that leads to harm and risks of harm to children. Therefore, your children’s risk assessment should not be limited to considering individual pieces of content, but rather should consider how your service is used overall.

Identify non-designated content for assessment

4.8 Non-designated content is set out in the Act as content “of a kind which presents a material risk of significant harm to an appreciable number of children in the United Kingdom”.45 All service providers completing a children’s risk assessment have a duty to assess the risk of children encountering non-designated content on their service.
4.9 Ofcom has identified two categories of non-designated content.46 You should consider the risk of children encountering these categories of non-designated content, and you should also consider whether there are other kinds of non-designated content that Ofcom has not identified in the Children’s Register that may be relevant to your service. We have included more detailed guidance on this in Part 3, in the ‘Non-designated content’ sub-section.

Consult Ofcom’s Children’s Risk Profiles

4.10 Service providers must take account of the relevant Children’s Risk Profiles when conducting their children’s risk assessment.47 There are separate Children’s Risk Profiles for user-to-user services and search services. The Children’s Risk Profiles identify risk factors associated with each kind of primary priority content, each kind of priority content, and for the kinds of non-designated content that Ofcom has identified.48 These risk factors are a good starting point when thinking about content harmful to children on your service, and are important to understand before carrying out the assessment at Step 2.
4.11 However, the risk factors are not comprehensive. You may find it helpful to consult the Children’s Register to gain a better understanding of each particular kind of content harmful to children, how it could manifest on your service and the risk factors associated with it.
4.12 You should use the risk factors identified when assessing content harmful to children in Step 2. See the sub-section ‘Children’s Risk Profiles’ in Part 3 for the Children’s Risk Profiles and how to use them.

Essential records for Step 1

4.13 You should keep written records confirming that you have consulted Ofcom’s Children’s Risk Profiles.
44 Children’s Register of Risks and Guidance on Content Harmful to Children 45 Section 60(2)(c) of the Act.
46 These are: content that discriminates against or otherwise stigmatises body types or body parts; and content that romanticises depression and depressive thinking. 47 Note that the separate Illegal Content Risk Assessment Guidance includes different Ofcom Risk Profiles to help meet your illegal content risk assessment duty. 48 The Children’s Risk Profiles have grouped different content harmful to children in line with the Children’s Register. Regardless of this, service providers have a duty to separately assess the risk of children encountering each kind of primary priority content, each kind of priority content, and non-designated content.
4.14 You should keep written records of all of your service’s relevant risk factors by using the check boxes provided in the Children’s Risk Profiles. Note that each risk factor will be relevant to one or more particular kinds of content harmful to children as explained in the Children’s Risk Profiles.
4.15 You must make a record of any kind(s) of non-designated content you identify for assessment on your service. You must also record how you have considered the risk of children encountering non-designated content by means of your service, including when you do not identify any kind(s) of non-designated content for assessment, beyond those identified by Ofcom.

Step 2: Assess the risk of harm to children

4.16 In Step 1, you will have understood each kind of primary priority content, each kind of priority content, and any kind(s) of non-designated content (including Ofcom identified non-designated content) that you need to assess for your service. You will also have used the Children’s Risk Profiles to identify the risk factors that relate to each particular kind of content harmful to children.49
4.17 The objective of this next step will be to use evidence to assess and assign a risk level for each of the four kinds of primary priority content, each of the eight kinds of priority content, and any kind(s) of non-designated content you have identified for assessment. When doing this, you must give separate consideration to children in different age groups, and consider the effect of design and use on the level of risk of harm.

Sequence of activities and outcomes for Step 2

4.18 You now need to assess the level of risk of harm presented by your service for each kind of content harmful to children identified in Step 1. As part of this, you should consider the likelihood and impact of children encountering each kind of primary priority content, each kind of priority content, and any kind(s) of non-designated content you have identified for assessment on your service.
4.19 This means that you should:
• Consider if there are any additional characteristics of your service which are not present in Ofcom’s Children’s Risk Profiles, but which might increase risk of harm to children. • Consider if there are any existing controls on your service which affect the level of risk of content harmful to children on your service. If so, you should consider how and to what extent these controls affect the risk of harm to children. • Identify and use evidence relating to your service to complete your children’s risk assessment. The level of risk of content harmful to children can be influenced by various elements, including how your service’s functionalities, user base (including the different age groups of children), business model, and systems and processes in combination can serve to increase or decrease risks to children.
49 Aside from kinds of non-designated content that you have identified beyond the types from Ofcom’s Children’s Register, which are not associated with Children’s Risk Profiles.
4.20 The children’s risk assessment should not be a theoretical exercise. You should use evidence to assess the ways the service is, or could be, actually used. We provide guidance on how this can be done in the sub-section ‘Identifying relevant evidence’.
4.21 To complete a ‘suitable and sufficient’ assessment, you must assess everything set out in the Act (sections 11(6) for user-to-user services and 28(5) for search services), and assign a risk level based on evidence about the likelihood and impact of children encountering each of the four kinds of primary priority content, each of the eight kinds of priority content, and any kind(s) of non-designated content that you have identified for assessment on your service.
4.22 To assist you with this, we have provided a Risk Level Table to help inform your judgement on the risk level for each kind of primary priority content, each kind of priority content, and any kind(s) of non-designated content that you identify for assessment on your service.50
4.23 When you are considering likelihood and impact, you must give separate consideration to how each kind of primary priority content, priority content and non-designated content present risks to children in different age groups.
4.24 You must also consider how the design and use of your service affects the level of risk of harm to children, in particular: 51 a) Features and functionalities which present higher levels of risk, which include, for user-to-user services, those which enable adults to search for or contact children; and, for search services, those which make suggestions to user’s searches (predictive search). b) Features and functionalities which impact how much children use the service, such as features which enable content to play automatically.

Identifying relevant evidence

4.25 You must use appropriate evidence to make accurate judgements about the level of risk of content harmful to children, as well as how the design and use of the service affects the level of risk.
4.26 Table 3 lists the two types of evidence inputs you should consider using: core inputs and enhanced inputs. All service providers should use all of the core inputs. Providers of large services, and service providers who identify several specific risk factors for a kind of content harmful to children using the Children’s Risk Profiles, should also consider using enhanced inputs. Providers of other services that have enhanced inputs available should also consider using them. The ‘Evidence Inputs’ sub-section in Part 3 provides more detailed information on these types of evidence.

Table 3: Summary of relevant types of evidence

Type
Overview of inputs
Core inputs
• Risk factors identified through relevant Children’s Risk Profile (Step 1). • User complaints and reports. • User data (e.g., age, language, groups at risk).
All service providers should
consider
• Available evidence relevant to understanding the age of children on your service. • Retrospective analysis of incidents of harm to children. • Relevant sections of Ofcom’s Children’s Register of Risks and Guidance on Content Harmful to Children.52 • Relevant findings of your illegal content risk assessments. • Relevant findings of your children’s access assessments. • Evidence drawn from existing controls.53 54 • Other relevant information (including other characteristics that apply to your service that may increase or decrease risks of harm to children).
Enhanced inputs
• Results of product testing. • Results of content moderation systems.55 • Consultation with internal experts on risks and technical mitigations. • Views of independent experts. • Internal and external commissioned research. • Outcomes of external audit or other risk assurance processes. • Consultation with users. • Results of engagement with relevant representative groups.
Should be considered by large
service providers and those who
have identified multiple specific
risk factors for a kind of content
harmful to children.
Should be considered by all
service providers where the core
inputs alone are not sufficient to
support an assessment of level
of risk.
Note: These are not exhaustive examples – may vary by service or business model. See also Part 3
‘Evidence Inputs’.
50 See ‘Risk Level Table for content harmful to children’ in Part 3 of this guidance. 51 Section 11(e)-(h) of the Act.
4.27 When considering user data, we encourage you to familiarise yourself with the data protection legislation, and to consult the Information Commissioner’s Office’s (ICO) guidance on UK GDPR requirements and the Children’s Code.56

Evaluate likelihood and impact by assigning a risk level to content harmful to children

4.28 You should use your judgement and consider relevant evidence to assign a risk level of high, medium, low or negligible to each of the four kinds of primary priority content, each of the eight kinds of priority content, and to any kind(s) of non-designated content that you have identified for assessment on your service. You should have at least two risk levels for non-designated content, and additional risk levels for any other kind(s) of non-designated
52 A service provider can consult the Children’s Register to better understand different kinds of content that is harmful to children, for instance, those which they identify risk factors for when consulting Children’s Risk Profiles. 53 This could include insights from content moderation systems – as a core input, we expect service providers to consider high level outputs, for example, the volume of content harmful to children identified through a complaints-based content moderation system.
54 This could also include results of previous interventions to reduce online safety risk. 55 As an enhanced input, we expect service providers to consider more granular or in-depth analysis such as evidence which relates to the effectiveness of content moderation systems. 56 ICO, UK GDPR guidance and resources; ICO, Age appropriate design: a code of practice for online services; ICO, ‘Likely to be accessed’ by children guidance.
content you have identified for assessment. We have included additional guidance on assessing non-designated content in Part 3.
4.29 The risk level that you assign to each particular kind of content harmful to children will be important when considering which safety measures you should implement as part of Step 3. Note that certain measures listed in the Codes address different particular kinds of content harmful to children.

Assessing the likelihood of children encountering harmful content

4.30 When evaluating likelihood, you should ask yourself the questions set out in Table 3.
Table 3: What to consider when assessing likelihood of children encountering content harmful to children Guiding questions when assessing likelihood • If your service is a user-to-user service, do your service’s risk factors identified in Step 1 indicate that children are likely to encounter this kind of content on your service?57 If so, how many risk factors do you have? Ordinarily, the larger the number of risk factors for a given kind of content, the higher the likelihood of that kind of content. If your service is a search service, does the Children’s Register section58 on search indicate evidence that children are likely to encounter this kind of content on your service? Is there evidence to indicate that children are likely to encounter any of the kind(s) of non-designated content you have identified for assessment on your service? If so, what is the incidence of such kinds? • Are there any additional characteristics of your service (including functionalities, child user base, business model and governance, systems and processes) that may make it more likely that children encounter this kind of content? For example, you may consider: > (For user-to-user services) Are adults able to search for children or contact children through your service? > (For search services) Do you provide a functionality that makes suggestions relating to child users’ search requests? > Does your service use recommender algorithms or functionalities which allow users to share content widely? > Do you have any features or functionalities which affect how much children use your service, such as a feature that enables content to play automatically?
Evaluating the likelihood will depend on your understanding of the evidence about your own service. • Is there any evidence from your core inputs that content harmful to children is likely to occur on your service? You should consider:
57 If you have identified additional kinds of non-designated content for assessment outside of those that Ofcom has identified, you should consider whether any risk factors you have identified for other content harmful to children are relevant to assess likelihood. 58 Section 12 of the Children’s Register.
> Evidence of this kind of content being encountered by children, including children with certain characteristics, based on user complaints and reports. For example, significant volumes of reports in relation to a particular kind of content could indicate a higher likelihood of that kind of content occurring; and > Any other relevant evidence and data which suggests there is a risk of this kind of content occurring on your service.
• If you have consulted core inputs and are still unsure about the likelihood of children encountering this kind of content, consider any additional evidence from enhanced inputs. For example, you may consider: > Evidence from independent experts or externally commissioned research that highlights the potential for children to encounter this kind of content; > Evidence based on results of product testing of the potential for children to encounter this kind of content; and > Evidence based on results of content moderation of this kind of content.
• Are there measures already in place that reduce the risk of children encountering this kind of content on your service? Can you demonstrate that these are effective in decreasing the risk of children encountering this kind of content? For example, you could consider: > Do you prohibit this kind of content on your service and do you have effective controls to identify and remove this content from the service? > Have you implemented highly effective age assurance to prevent/protect children from encountering this content? > Do you have measures to promote users’ media literacy and safe use of the service? It is possible that such measures could reduce the likelihood of content harmful to children being encountered.
Source: Ofcom analysis

Assessing the impact of children encountering harmful content

4.31 When evaluating the impact of children encountering each kind of primary priority content, each kind of priority content, and any kind(s) of non-designated content you have identified for assessment, you should ask yourself the questions set out in Table 4.
Table 4: What to consider when assessing the impact of children encountering content harmful to children Guiding questions when assessing impact • To make judgements on the nature and severity of content harmful to children, you need to consider: > If children on your service have had a materially harmful experience, for example, due to the nature of this kind of content and how children may encounter it on the service. > If harm is suffered indirectly by children who are not users of the service, for example, violent content that normalises the carrying of weapons, by which other children are impacted.59 If so, how severe is the impact likely to be?
59 See Section 7 of Children’s Register.
> What the potential reach of this kind of content on your service could be, and the number of children that could be impacted. > How children might be affected by experiencing this kind of content cumulatively over time on your service?
• To make judgements on impact on the children affected, you need to consider: > How many children are on your service, either confirmed or estimated with your best available information?60 > What does child user data that you have available tell you about your user base demographics (including age, sex and any vulnerable groups)? > How many of your child users are particularly likely to be affected by the kind of content in question due to their characteristics, such as age, or belonging to vulnerable groups. > How are children who are not users of your service affected?
• To make judgements about the design and use of your service, you need to consider: > How does your service’s revenue model and commercial profile influence the way this kind of content is experienced on your service? Consider the information provided in the Children’s Risk Profiles and your own evidence. > Are there any other characteristics that apply to your service (including functionalities, user-base, business model and governance, and systems and processes) that you have identified may increase the impact of this kind of content? > Whether the way this kind of content is shared and disseminated, including through recommender systems and other algorithmic systems, could increase the number of children encountering this kind of content over a period. > Whether you have any features or functionalities which affect how much children will use the service. The more time a child spends on a service the greater the impact of encountering this kind of content may be.
Evaluating the impact will depend on your understanding of the evidence about your own service.
• Is there any evidence from core inputs about the experience of this kind of content and its impact? For example: > What user complaints and reports regarding this kind of content tells you about impact on child users and other children. > The potential reach of this kind of content measured by the number of child users (if you have this evidence available) who could be affected.
• If you have consulted core inputs and are still unsure about the impact of the content, then consider any additional evidence based on the information from enhanced inputs. For example, you may consider: > What user research, including with children (including children in different age groups, or children belonging to vulnerable groups) shows about impact on child users and other children;
60 Please see the ‘Assessing the user base’ sub-section in this guidance document.
> What independent experts (e.g., child behavioural experts) or research tells you about the impact of this kind of content on a service of your type; > Identifying metrics regarding the virality of this kind of content (including its potential reach and speed of spread); and > Evidence about how this kind of content may affect third parties beyond your service.
• For the avoidance of doubt, where evidence shows the potential for severe harm in relation to a kind of content, we expect that this may lead to an assessment of medium or high impact, even if the number of children potentially impacted is relatively small or smaller than the indicative values provided in the Risk Level Table (Risk Level Table included in Part 3 of this guidance). Source: Ofcom analysis

Assessing the user base

4.32 When assigning risk levels to content harmful to children, you must give separate consideration to how each particular kind of content harmful to children presents risks to children in different age groups, and consider the effect of design and use on the level of risk.
4.33 You must assess your service’s user base to conduct a suitable and sufficient children’s risk assessment. We expect this to reflect understanding of how many users are children and assessing if child users are in different age groups. We recognise service providers may have different levels of understanding, evidence and data about the age or age range of their child users, and provide guidance on this in the sub-section ‘Evidence on child age’ in Part 3.
4.34 We recognise that, in many cases, service providers will be unable to determine the age ranges of their users (whether differentiating children from adults; or children within different age groups) to a high level of confidence. As such, we expect children’s risk assessments to err on the side of caution when assessing the scope for children to be impacted by harmful content. For example, if there is a high number of overall users on your service, we expect you to assume that you also have a high number of children, potentially in younger age groups, that are particularly vulnerable to harm from the relevant kind of content, unless you have appropriate evidence to demonstrate otherwise.

Essential records for Step 2

4.35 Where applicable, you should keep a list of any additional characteristics (including user base, business models, functionalities, governance, and systems and processes) considered alongside the risk factors identified in Ofcom’s Children’s Risk Profiles. This should include how you have identified and assessed those functionalities that present higher levels of risk, such as recommender systems, functionalities which enable adults to search for and/or contact children and predictive search functionalities, as well as features and functionalities which affect how much children use the service.
4.36 If you have considered the role of any existing controls already in operation on your service at the time of this risk assessment, you should record what these controls are, what risks they are intended to mitigate and how they do this, and how the consideration of the existing controls has impacted the risk level you have assigned to a kind of content harmful to children.
4.37 You should keep a list of the evidence and summary of the reasoning that has informed the assessment of likelihood and impact for each kind of primary priority content, each kind of priority content, and any kind(s) of non-designated content you have identified for assessment, and how this has informed the risk level assigned to each. This should include relevant evidence about the design and use of the service, and how different age groups of children have been considered.
4.38 You should keep written records of the level of risk assigned to each of the four kinds of primary priority content, each of the eight kinds of priority content, and for any kind(s) of non-designated content you have assessed on your service, together with an evidence-based explanation of the decision. This level should reflect risk as it exists on the service at the time of assessment.

Step 3: Decide measures, implement and record

4.39 In Step 2, you will have assigned a risk level to each of the four kinds of primary priority content, each of the eight kinds of priority content, and any kind(s) of non-designated content you have identified for assessment on your service. You now need to decide how you will meet the children’s safety duties. The objective of this step is to make a record of your assessment, identify any relevant measures to implement to address risk to children, and record any measures you have taken.

Sequence of activities and outcomes for Step 3

Decide what measures you should take to reduce the risk of harm

4.40 You should now refer to Ofcom’s Protection of Children Codes.61 The measures recommended for your service will be based on your service’s size, functionalities or risk levels: a) Some measures in the Protection of Children Codes are applicable to all services. b) Other measures in the Protection of Children Codes will be informed by your assignment of risk levels in Step 2. Some of the measures only apply to services which are medium or high risk for one particular kind of content harmful to children. Some measures in the Protection of Children Codes only apply to services which are medium or high risk for two or more particular kinds of primary priority content, priority content or non-designated content. c) The size of your service, or other functionalities and characteristics of your service will also determine whether certain measures in the Protection of Children Codes are recommended for your service.62
4.41 The Act says the Codes are like a ‘safe harbour’, meaning that service providers who choose to implement all applicable measures in the Codes will be treated as complying with their relevant duties under the Act.63

Alternative measures to consider

4.42 You do not have to follow our Protection of Children Codes, and you may use alternative measures to comply with your duties to address the risks to children identified for your
61 Volume 4: ‘What should services do to mitigate the risks of online harms to children?’ 62 Section 10: ‘Framework for Codes’. 63 Section 49 of the Act.
service. If you do take alternative measures, you must keep a record of what you have done and explain how the relevant children’s safety duties have been met. In doing so, you must consider the importance of protecting users’ rights to freedom of expression and of protecting child users from breaches of relevant privacy laws.

Additional measures to consider

4.43 For some services, the measures recommended by our Protection of Children Codes may not eliminate all of the risks to children that have been identified. In such cases, we encourage you to consider additional measures to manage or mitigate your risks. For example, you may put in place bespoke controls needed for your unique risk context.

Implement all measures to mitigate and manage risk

4.44 Once you have decided which Protection of Children Codes measures to take, you should implement them.

Changes to measures and controls

4.45 If you make any changes to existing systems, processes or other measures in place on your service, this could affect your risk levels for each kind of primary priority content, each kind of priority content, and for any kind(s) of non-designated content you have identified for assessment, and you may need to re-evaluate if your approach is still adequate. If your existing measures contribute to reducing your risk level, we consider that it will be appropriate to continue implementing these, even if they are additional or alternative to those recommended in our Protection of Children Codes.
4.46 If you stop implementing existing measures, this may constitute a significant change and may impact your risk levels. Further guidance on what constitutes a significant change is set out in Part 3.

Essential records for Step 3

Record the outcomes of the children’s risk assessment

4.47 At this stage, you should ensure you have a written record of the findings of your children’s risk assessment, also including details about how it has been carried out. Records should be durable, accessible, easy to understand and up to date, as set out in our Record-Keeping and Review Guidance.
4.48 You should keep a record of all measures from Ofcom’s Protection of Children’s Codes that have been, or are planned to be, implemented.
4.49 If you have chosen to take alternative measures, you must record:
• The applicable measures that are not planned to be implemented; • The alternative measures that have been or are planned to be implemented instead; • Information that demonstrates how these measures achieve compliance with the safety duties.
4.50 Further information is provided in section 5 of the Record-Keeping and Review Guidance (‘Records of alternative measures taken to comply with a relevant duty’).
4.51 Well-maintained and accurate records and regular, timely reviews of compliance will help you to keep track of how you are complying with the children’s safety duties and ensure that the measures that you have taken are appropriate.
4.52 When making a record of your children’s risk assessment, you should capture all information shown in Box 2. This will help you to ensure that your children’s risk assessment is suitable and sufficient, that you have considered all the elements of section 11 or section 28 of the Act (as applicable), and have included the evidence you have relied on to assess the risks relevant to your service.
Box 2: Information to include in the record of your children’s risk assessment • The service to which the children’s risk assessment relates; • The date the children’s risk assessment was completed; • If applicable, the date the children’s risk assessment was reviewed or updated; • Who completed the children’s risk assessment, and the named person responsible for the children’s risk assessment; • Who approved the children’s risk assessment; • Confirmation that you have consulted Ofcom’s Children’s Risk Profiles. You may do this by recording the outcomes of the Children’s Risk Profiles questionnaire, see Part 3. • A record of any risk factors from Ofcom’s Children’s Risk Profiles which are relevant to your service; • A record of the kind(s) of non-designated content that you have identified for assessment, and how you have considered the risk of children encountering non-designated content by means of your service; • Where applicable, a list of any additional characteristics (including user base, business models, functionalities, governance, and systems and processes) considered alongside the risk factors identified in Ofcom’s Children’s Risk Profiles in Step 1. This should include how you have identified and assessed those functionalities that present higher levels of risk, such as recommender systems, functionalities which enable adults to search for and/or contact children and predicting search functionalities, as well as features and functionalities which affect how much children use the service; • If you have considered the role of any existing controls already in operation on your service at the time of this risk assessment, you should record what these controls are, what risks they are intended to mitigate and how they do this, and how the consideration of the existing controls has impacted the risk level you have assigned to a kind of content harmful to children; • A list of the evidence and summary of the reasoning that has informed the assessment of likelihood and impact of each kind of primary priority content, each kind of priority content, and each kind of non-designated content. This should include relevant evidence about the design and use of the service, and how different age groups of children have been considered; • The level of risk assigned to each of the four kinds of primary priority content, each of the eight kinds of priority content, and for any kind(s) of non-designated content you have assessed on your service, and an evidence-based explanation of the decision; • Confirmation that the findings of the children’s risk assessment have been reported, and recorded, through appropriate governance and accountability channels (this is explained in Step 4); and
• Information regarding how your service takes appropriate steps to keep the children’s risk assessment up to date (e.g., a written policy) (this is explained in Step 4).

Record all relevant measures and how the safety duties have been met

4.53 You have identified, implemented and made a written record of all measures from Ofcom’s Protection of Children Codes that have been, or are planned to be, implemented.
4.54 You have made written records of the measures that are recommended in the Codes for your service, but not planned to be implemented, and the measures that are alternatives to those set out in Ofcom’s Protection of Children Codes, with information that demonstrates how these alternative measures meet the relevant duties.

Step 4: Report, review and update

4.55 The objective of this step is to understand how to keep your children’s risk assessment up to date and reported appropriately, both internally and externally.
4.56 In Step 3, you will have made a complete record of your children’s risk assessment, identified the relevant codes measures to implement to address your identified risks, and recorded any alternative measures you have taken and explained how they meet the relevant safety duties, as well as which codes measures they are alternatives to.
4.57 In this final step, you will understand and implement the systems and processes by which your children’s risk assessment should be reviewed and updated, as well as implementing arrangements to monitor the effectiveness of your implemented safety measures from Step 3.

Sequence of activities and outcomes for Step 4

Report on the children’s risk assessment and measures via relevant governance and accountability channels

4.58 If you provide a service that is likely to be accessed by children, you must keep a written record of its children’s risk assessment, including details of how the risk assessment was carried out and its findings.
4.59 Reporting on risk is a primary element of good practice in risk management. Accurate and timely reporting through appropriate governance channels improves organisational oversight of risk and leads to better risk management outcomes.
4.60 Our Protection of Children Codes include specific Governance and Accountability measures to help you meet their separate safety duty to mitigate and manage risk.64 Having in place adequate governance is key to how you identify, assess, manage, report and review online safety risks to children to meet your children’s risk assessment duties. This also will help you to consider the range of risks identified in your latest children’s risk assessment.
4.61 If you provide a Category 1 or 2A service, you must also supply Ofcom with a copy of your children’s risk assessment as soon as reasonably practicable after completing or revising your children’s risk assessment. Category 1 service providers must also include a summary
64 PCU/PCS A1-A7.
of the findings of their children’s risk assessment in their terms of service, and Category 2A service providers must include a summary of the findings of their children’s risk assessment in a publicly available statement.

Notify Ofcom of the kinds and incidence of non-designated content

4.62 Where you provide a user-to-user service and you identify that non-designated content is present on your service, then you must notify Ofcom of:65 a) The kinds of content identified as non-designated content; and b) The incidence of those kinds of content on your service.
4.63 You should notify Ofcom using the following email address: nondesignatedcontent@ofcom.org.uk.
4.64 We have included more detailed guidance on how to do this in Part 3, sub-section ‘Non-designated content’.

Monitor the effectiveness of your safety measures

4.65 Monitoring the effectiveness of the measures you implement, alongside your levels of residual risk, is important for ongoing risk management.66 This will also help you keep your children’s risk assessment up to date, as we explain in Part 1, sub-section ‘When to review or carry out a new children’s risk assessment’.

Review your children’s risk assessment

4.66 You will need to keep your children’s risk assessment up to date by reviewing your assessment at least annually. We explain this in Part 1, sub-section ‘When to review or carry out a new children’s risk assessment’. A review involves checking that your latest risk assessment still accurately reflects the risks on your service. If there have been very few or minor changes to the design, operation and user base of your service since your last risk assessment, you may consider a limited review to be adequate.67
4.67 Outside of annual reviews, the Act includes triggers for service providers to: a) Review your assessment if Ofcom makes a significant change to a Children’s Risk Profile relevant to your service; b) Before making a significant change to the design or operation of your service, a duty to carry out a new children’s risk assessment in relation to this change.68
65 Whether this is non-designated content which Ofcom has identified in the Children’s Register, or any other kinds of non-designated content which Ofcom has not identified in the Children’s Register. 66 Governance and Accountability measure PCU A4 in the Code for user-to-user services/PCS A4 in the Code for search services complements this. This measure recommends that providers of all large user-to-user services and all large search services likely to be accessed by children that are multi-risk for content harmful to children should have an internal monitoring and assurance function to independently assess, on an ongoing basis, the effectiveness of measures to mitigate and manage risks of harm to children identified in the children’s risk assessment.
67 Governance and Accountability measure PCU A1 in the Code for user-to-user services/PCS A1 in the Code for search services complements this. The measure says that a provider’s most senior governance body in relation to the service should carry out and record an annual review of risk management activities having to do with content that is harmful to children in the UK, including in relation to risk that is remaining after the implementation of appropriate Codes measures. 68 This could include deciding to change any existing measures, or measures which are additional to those set out in the Codes.
4.68 Further guidance on what constitutes a significant change is in Part 3 of this guidance under sub-section ‘Making a significant change to your service’.

Essential records for Step 4

4.69 You will have made a written record of the annual review cycle for your children’s risk assessment, and the named person responsible who has been appointed for this process.
4.70 You will, if you are a provider of a Category 1 or 2A service, have supplied Ofcom with a copy of your children’s risk assessment record made in Step 3.
4.71 You will, if you are a provider of a Category 1 service, have summarised the findings of your most recent children’s risk assessment in your terms of service.
4.72 You will, if you are a provider of a Category 2A service, have summarised the findings of your most recent children’s risk assessment in a publicly available statement.
4.73 You will have made a written record which confirms that the findings of the children’s risk assessment have been reported, and recorded, through appropriate governance and accountability channels.69
69 Service providers should identify the relevant Governance and Accountability measures in the Protection of Children Codes of Practice that may apply to them.

Part 3: Supporting documents

Section 1 Children’s Risk Profiles

1.1 The Children’s Risk Profiles are resources to consult when conducting your children’s risk assessment. All service providers that are subject to the children’s risk assessment duties must take account of the relevant Children’s Risk Profiles when conducting their own children’s risk assessment.
1.2 The Children’s Risk Profiles are made up of a list of different risk factors. After consulting the Children’s Risk Profiles (using our guidance), you should have a list of risk factors and the kinds of content harmful to children most strongly associated with these risk factors relevant to your service.70 You should then assess these risk factors alongside your own evidence under Step 2.
1.3 These risk factors represent a selection of service characteristics (such as user base, business models and functionalities) that our Children’s Register of Risks (Children’s Register) indicates are most strongly linked to a risk of one or more kinds of content harmful to children outlined in the Children’s Register.71
1.4 The Children’s Register provides a detailed analysis of the risks of harm to children posed by content harmful to children that we have identified across user-to-user and search services. It therefore contains our evidence in full and in some cases identifies risk factors in addition to those highlighted in the Children’s Risk Profiles. The Children’s Risk Profiles will be updated as necessary when changes are made to the Children’s Register.
1.5 When consulting the list of risk factors in the Children’s Risk Profiles, you should keep in mind: • We do not include all the characteristics that may lead to a risk of harm. For example, we do not always include risk factors from the Children’s Register where we have more limited evidence, or where we have drawn parallels based on the similarity between two kinds of content harmful to children. • The description of the risks provided is a high-level summary only. The effect of any risk factors will vary depending on the context, including the combinations of risk factors present, the governance, systems and processes a service has in place, and the dynamics that may be unique to the kind of content harmful to children or the nature of a service itself.72
1.6 You should see the Children’s Risk Profiles as the starting point of the four-step children’s risk assessment process. They will help you to understand which kinds of content harmful to children are most likely to be encountered by children on a service like yours, and which risk factors are most likely to play a role.
70 The key kinds of content harmful to children associated with a risk factor are those where our evidence indicated the strongest link. There may be other kinds of content harmful to children which may be relevant. For further information, see Section 1 of the Children’s Register. 71 Table 1.1 and 1.2 in Section 1 of the Children’s Register. 72 For further information on how we see these dynamics play out in our evidence base, see Section 1 and Section 16 of the Children’s Register.

Children’s User-to-User Risk Profile

1.7 The Children’s User-to-User Risk Profile is presented in Table 5. Each row represents a unique risk factor that services should consider when conducting their children’s risk assessment. The information provided on the risk factors in the table is based on the evidence in the Children’s Register.
1.8 When consulting the table, you should do the following: • First, answer the ‘Yes’/‘No’ questions in Figure [] about the characteristics of your service.73 • Second, use your answers to select which specific risk factors from Table 5 apply to you. Each ‘Yes’ answer corresponds to a risk factor you will need to take account of in your children’s risk assessment. For example, if you answered ‘Yes’ to questions 2a, 3a and 6, then you should select User profiles, User connections and Content recommender systems from the Children’s User-to-User Risk Profile. A Glossary is available to help you interpret your risk factors.74 • Third, review the four general risk factors at the bottom of the Children’s User-to-User Risk Profile: user base age, other user base demographics, business model and commercial profile. These apply to all services, and you will need to take account of each in your children’s risk assessment.
1.9 After you have taken these three steps, you should have the list of risk factors you will need to take account of when conducting your own children’s risk assessment. This list includes any specific risk factors you have selected, plus all four of the general risk factors.
1.10 Step 2 of the Children’s Risk Assessment Guidance provides details on how to use this list of risk factors as part of your children’s risk assessment. At Step 2, you will also consider how the risk factors you have selected affect your service (e.g., whether this is a risk that you are already managing, or one that you may need to address through a set of mitigations to help reduce the risk of harm to children).
1.11 As part of Step 2, you will be asked to assess the impact of content harmful to children on your service, which will include assigning risk levels to each kind of content harmful to children. We explain in the Children’s Risk Assessment Guidance that if you have identified a higher number of risk factors for each type of content, there may be a higher likelihood of children encountering this kind of content on your service. To help you work out the proportion of risk factors you have identified, we produced Table 6 for service providers to consult when they are using the Risk Level Table provided in the Children’s Risk Assessment Guidance. To clarify, you are not legally required to take account of this table, as you need to do with the Children’s User-to-User Risk Profile, but we consider that this is a helpful way to understand which risk factors are associated with each type of harmful content and the proportion that are relevant to you.
73 If your service offers multiple versions – for example, mobile and web – you should select ‘Y’ if any version of the service has the relevant characteristic(s). However, this only applies where versions are similar enough to be treated as a single service. 74 If, after consulting Section 18 of the Children’s Register, you are still unsure if the risk factor applies to you, we would suggest you read the corresponding information provided about that risk factor in Table 7 and consider if this information is relevant to your service. You may also wish to consult the Children’s Register for more detailed information on the corresponding risk factor or kind of content harmful to children.
Figure 1. Questions for identifying your risk factors
Select Yes (Y) or No (N) for the following questions about your user-to-user service.
1. Is my service any of the following service types? Select all that apply:75
Y / N Y / N Y / N Y / N Y / N Y / N
a. Social media services (services which connect users and enable them to build
communities around common interests or connections).
b. Messaging services (services that are typically centred around allowing users to
send messages that can only be viewed or read by a specific recipient or group of
people).
c. Gaming services (services that allow users to interact within partially or fully
simulated virtual environments).
d. Pornography services (services whose principal purpose is the hosting or
dissemination of pornographic content and who host user-generated pornographic
content).
e. Discussion forums and chat room services (services which allow users to send or
post messages that can be read by the public or an open group of people).
f. Video-sharing services (services that allow users to upload and share videos with
the public).
2. Does my service have any of the following functionalities related to how users identify themselves to one another? Select all that apply: a. Users can display identifying information through a user profile that can be viewed by others (e.g., images, usernames, age). b. Users can share content anonymously (e.g., anonymous profiles or access without an account).
Y / N Y / N
3. Does my service have any of the following functionalities related to how users network with one another? Select all that apply: a. Users can connect with other users.76 b. Users can form groups or send group messages.
Y / N
Y / N
4. Does my service have any of the following functionalities that allow users to communicate with one another? Select all that apply: a. Livestreaming. b. Direct messaging. c. Commenting on content. d. Posting images and videos. e. Re-posting or forwarding content.
Y / N
Y / N
Y / N
Y / N
Y / N
5. Does my service have any of the following functionalities that allow users to find or encounter content? Select all that apply: a. Searching for user-generated content. b. Content tagging.
Y / N Y / N
6. Does my service use content recommender systems?
Y / N
7. Does my service have functionalities and other features that increase user engagement?
Y / N
(e.g., infinite scrolling and autoplay features, notifications and alerts)
75 A service may consider more than one service type to apply. 76 We describe ‘user connections’ as a user-to-user service functionality that allows users to follow or subscribe to other users. Users must sometimes be connected to view all or some of the content that each user shares. Further information on risk factors is available in Section 18 of the Children’s Register.

Table 5. Children’s User-to-User Risk Profile77

Specific risk factors user-to-user services with relevant characteristics should take account of in
their children’s risk assessment
1. Service type factors
1a Social media services
• Risk factor: Social media services. • Key kinds of content harmful to children:* Your service is likely to have an increased risk for all kinds of content harmful to children. Many social media services are designed to maximise engagement between users and with content. They also typically enable users to connect and create communities with other users. If your service is a social media service, you should consider how this can increase the risk of children encountering content harmful to children on your service. Research shows that social media services can increase the risk of children encountering all kinds of primary priority and priority content that is harmful to children, and both kinds of non-designated content outlined in Ofcom’s Children’s Register.
1b Messaging services
• Risk factor: Messaging services. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to bullying content and violent content. Messaging services allow users to protect their privacy. If your service is a messaging service, you should consider how this can play a role in children encountering content harmful to children in a setting that is hidden from public view. Our evidence indicates that this can result in more targeted harmful behaviour, such as bullying content, and make detection more difficult. This can be particularly true for messaging services with encrypted and ephemeral messaging, which may be more likely to be used by children sharing violent content.
1c Discussion forums and chat room services
• Risk factor: Discussion forums and chat room services. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to eating disorder content as well as suicide and self-harm content. If your service is a discussion forum or chat room, you should consider how your service may be used to discuss and share content harmful to children in a setting that is typically visible to the public. For example, our evidence shows that these services can act as spaces where suicide and self-harm is encouraged and eating disorder content is shared among dedicated communities.
1d Gaming services
• Risk factor: Gaming services. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to abuse and hate content, bullying content and violent content. If your service is a gaming service, you should consider how it may bring children in contact with users who share content harmful to children and result in harmful interactions. For instance, our evidence shows that children can often be bullied by users on these services. Children can also be excluded from games due to their identity and receive hateful messages.
Specific risk factors user-to-user services with relevant characteristics should take account of in
their children’s risk assessment
1e Pornography services
• Risk factor: Pornography services. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to pornographic content. If your service is a pornography service, you should consider how your service may be used by children to access pornographic content. While pornographic content can be encountered by children on a range of services, our evidence shows that pornography services are a key space in which children can search for and view pornographic content.
1f Video-sharing services
• Risk factor: Video-sharing services. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to abuse and hate content, content that shames or otherwise stigmatises body types or physical features, content that promotes depression, hopelessness and despair, dangerous stunts and challenges content, eating disorder content, harmful substances content, suicide and self-harm content, and violent content. If your service is a video-sharing service, you should consider how your service may be used to disseminate harmful video content that can be encountered by children. For instance, our evidence indicates that abuse and hate content, and violent content can often be encountered by children on video-sharing services. Furthermore, content encouraging eating disorders, body stigma content and depression content can also be recommended to users that may include children.
2. User identification factors
2a Services with user profiles
• Risk factor: User profiles. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to abuse and hate content and eating disorder content. If your service allows users to create a user profile that displays identifying information that can be viewed by others (e.g., images, usernames, age), we expect you to take account of the risks that can arise from this. For example, our evidence indicates that children can see users (e.g., ‘influencers’) with a significant number of user connections (see 3a) displayed on their user profiles as trusted sources of information, creating a heightened risk of harm if abuse and hate content or eating disorder content is shared from these accounts. Children can also display pro-eating disorder information on their user profile, such as in their username, which can contribute to them receiving recommendations for eating disorder content and facilitate the creation of networks with other pro-eating disorder users.
2b Services where users can post or send content anonymously, including
• Risk factor: Anonymous user profiles78 or users without accounts. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to abuse and hate content, bullying content, content that promotes depression, hopelessness and despair and violent content. Anonymity is an important tool for users (including children) to protect themselves from being identified and targeted for their views, particularly for marginalised communities and dissenting voices. However, our evidence indicates that if your
Specific risk factors user-to-user services with relevant characteristics should take account of in
their children’s risk assessment
without an account
service allows users to share content anonymously, risks can increase. Our evidence suggests these risks arise from the disinhibition effect, where users (including children) are emboldened because they cannot be identified by other users. For example, children can feel more encouraged to share violent as they assume their actions will not be traced back to them and often encounter violent content through anonymous ‘invite-only’ profiles that are dedicated to sharing such content. Anonymity can also make encountering content harmful to children such as bullying content more impactful, as not knowing who was involved can exacerbate the distress felt by children. Furthermore, depression content is often shared on pseudonymised accounts.
3. User networking factors
3a Services with user connections
• Risk factor: User connections. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to dangerous stunts and challenges content, eating disorder content, pornographic content, suicide and self-harm content, and violent content. User connections can allow children to build networks and help bolster the credibility of users that share content harmful to children. They can also play a role in children being contacted as users sometimes must be connected to message one another. Our evidence shows that children can often be sent pornographic content by users who they are connected with but do not know. Children can also be connected to users that are known to share content harmful to children such as self-harm content and eating disorder content. This makes such content easier to find and helps create networks of users in which content harmful to children such as online challenges and violent content can go ‘viral’, especially when users are connected in large numbers. As mentioned in user profiles (see 2a), children can perceive accounts with a large number of connections displayed on their profile as an indicator of trust.
Specific risk factors user-to-user services with relevant characteristics should take account of in
their children’s risk assessment
3b Services where users can form user groups or send group messages
• Risk factors: User groups. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to eating disorder content as well as suicide and self-harm content. User groups can enable users to create communities in which content harmful to children is shared and encountered by children. Our evidence shows that dedicated user groups are a key space in which pro-eating disorder communities can be formed and where harmful behaviours are encouraged. Children with lived experience of eating disorders, or who may be vulnerable to eating disorder content, are particularly likely to seek out eating disorder content through such groups. These groups can also act as spaces where users encourage others to seek support for suicide and self-harm, but also encourage others to engage in harmful behaviours. User groups dedicated to sharing suicide and self-harm content are sometimes reported to be ‘self-regulating’,79 with little perceived outside moderation, and therefore it is perceived to be easier to find or encounter content harmful to children on them. • Risk factor: Group messaging. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to abuse and hate content, bullying content, eating disorder content, pornographic content, suicide and self-harm content, and violent content. Group messaging can allow content harmful to children to be shared with multiple children at once in a more closed setting. Similar to user groups, children can find a sense of community in these spaces that may play a role in encouraging eating disorder content as well as promoting suicide and self-harm content. They can also act as spaces where children are targeted or witness the sharing of bullying content. Children can sometimes be added to often large group chats without giving permission or with users who they do not know who can share pornographic and violent content unexpectedly. This content can sometimes be more extreme in nature than content shared in open channels.
4. User communication factors
4a Services with livestreaming
• Risk factor: Livestreaming. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to suicide and self-harm content and violent content. Livestreaming can allow content harmful to children to be shared in real-time, which may be challenging to moderate. Our evidence shows that livestreaming can be used to share violent content and suicide and self-harm content that children can be exposed to. Livestreaming can intersect with group messaging (see 3b) and commenting (see 4c) functionalities to increase the risk of harm. While some users may use these messages or comments to express sympathy or coordinate help within the livestream, some can encourage suicide or serious self-harm.
Specific risk factors user-to-user services with relevant characteristics should take account of in
their children’s risk assessment
4b Services with direct messaging
• Risk factor: Direct messaging. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to abuse and hate content, bullying content and pornographic content. Direct messaging can allow users to share content harmful to children in a closed and more targeted manner. While direct messaging can enable users to protect their privacy, our evidence shows direct messaging can enable abuse and hate content, and bullying content behaviours, particularly between two users, that are more likely to go unnoticed by others. This risk may increase when users are able to message other users without the recipient’s permission. Children can also receive direct messages containing pornographic content, often in the form of hyperlinks and frequently by users they do not know or suspect to be ‘bots’.80
4c Services that enable commenting on content
• Risk factor: Commenting on content. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to abuse and hate content, bullying content, content that shames or otherwise stigmatises body types or physical features, eating disorder content, and suicide and self-harm content. Commenting on content can allow users to target or encourage children who share content harmful to children. For instance, hateful content or body shaming is often found in comments that can be encountered by children. Our evidence shows that children can be subjected to bullying content in the comments on posts they share, particularly when users can comment on content anonymously (see 2b). Harmful conversations that encourage eating disorder content can also take place in comments, which can cause children to further engage with this content and contribute to users contacting them in order to share more of such content due to the public visibility of their comments.
4d Services that enable posting images or videos
• Risk factor: Posting images or videos. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to abuse and hate content, content that promotes depression, hopelessness and despair, dangerous stunts and challenges content, eating disorder content, harmful substances content, pornographic content, suicide and self-harm content and violent content Posting images or videos can allow content harmful to children to be shared in more public settings with many users, including children. This can increase the likelihood of children encountering abuse and hate content, as well as depressive content, in more public settings. Online challenges and violent content are often posted in videos which have the potential to reach large audiences of children. This is also true of images depicting bodies which may be ‘aspirational’ to children with an eating disorder, particularly when posted from accounts with many user connections (see 3a) or found through content recommender systems (see 6) which can reach a large network of users. The ability to post images and videos ephemerally, such as on ‘Stories’, can also increase the risk of children sharing and encountering pornography and more extreme forms of violent content due to the perception that the content will be harder to moderate.
Specific risk factors user-to-user services with relevant characteristics should take account of in
their children’s risk assessment
4e Services that enable re-posting or forwarding content
• Risk factor: Re-posting or forwarding content. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to bullying content, suicide and self-harm content, and violent content. Re-posting or forwarding content can allow content harmful to children to be shared with a larger audience than it was originally shared with, often without the permission of the children that are featured in or targeted by this content, such as bullying content. It can also increase the likelihood of children encountering violent content that may be shared by users in their network. Engagement through reposting and reacting to content such as suicide and self-harm content can also be perceived as an indication of validation from other users on a service, which could encourage the further sharing of content harmful to children. This can be especially true when users who have a large number of user connections (see 3a) repost or react to content harmful to children posted by a child.
5. Content exploring factors
5a Services where users can search for user-generated content
• Risk factor: User-generated content searching. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to eating disorder content, pornographic content, and suicide and self-harm content. The ability to search for user-generated content within services may allow children to find content harmful to children on your service. For example, our evidence shows that children can make searches using code words to bypass content bans and support prompts in order to access eating disorder content. This may be more common among children with experience of eating disorders. Children may also make searches using hashtags (see content tagging, 5b) that are associated with content harmful to children such as suicide and self-harm content.
5b Services where users can tag content
• Risk factor: Content tagging. • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to content that promotes depression, hopelessness and despair, content that shames or otherwise stigmatises body types or physical features, dangerous stunts and challenges content, eating disorder content, harmful substances content, and suicide and self-harm content. Content tagging can improve the discoverability of content harmful to children and therefore make it more likely that such content will be encountered by children. For instance, content tagging can be used to direct users, including children, to content promoting the ingestion of harmful substances, dangerous stunts and challenges content by using certain hashtags. These hashtags can also direct children to body stigma and depression content. Our evidence also shows that users can create coded hashtags to obscure or disguise eating disorder content, as well as suicide and self-harm content in an attempt to evade content moderation practices. These hashtags can sometimes be used to disguise harmful content as recovery content or can appear alongside seemingly harmless hashtags, which can result in children encountering this content unexpectedly. There is also a risk of this content being mislabelled by content recommender systems (see 6) as a result and disseminated to other users.
Specific risk factors user-to-user services with relevant characteristics should take account of in
their children’s risk assessment
6. Recommender systems
6 Services with content recommender systems
• Risk factor: Content recommender systems • Key kinds of content harmful to children:* Your service is likely to have an increased risk of harm related to abuse and hate content, content that promotes depression, hopelessness and despair, content that shames or otherwise stigmatises body types or physical features, dangerous stunts and challenges content, eating disorder content, pornographic content, suicide and self-harm content, and violent content. Content recommender systems curate personalised content recommendations for users based on their profile or personal characteristics (e.g., age, location, gender) and their engagement patterns (e.g., viewing habits, reactions, reposts, comments). There is a risk of content recommender systems amplifying content harmful to children to a wide set of users, including children, who may otherwise not organically or intentionally come across this content. Our evidence indicates that content recommender systems that are primarily optimised for user engagement can suggest content harmful to children, such as violent content, to children without them actively seeking this content out. This risk is particularly pronounced when users create misleading content tags (see 5b) or keywords, such as hashtags that have been associated with content harmful to children. Children can also be exposed to content harmful to children after engaging with thematically similar content. For instance, they may be recommended suicide and self-harm content, body stigma content, and depression content after engaging with content relating to mental health. Children that have repeatedly engaged with content harmful to children can be recommended more of it with fewer alternative types of content, thereby increasing the risk of cumulative harm. Our evidence also indicates that in some cases children can over time be recommended harmful content that is associated with more severe impacts. For example, evidence reports children being presented with pornographic content depicting themes of violence after having engaged with pornographic content.
7. Other features
Specific risk factors user-to-user services with relevant characteristics should take account of in
their children’s risk assessment
7 Services with features that increase engagement
• Risk factor: Functionalities and other features that increase engagement. • Key kinds of content harmful to children:* Your service is likely to have an increased risk for all kinds of content harmful to children. Some functionalities and other features of a service affect how much users, including children, use the service. We have found that certain functionalities and other features on user-to-user services can cause children to spend more time using a service, and that high usage of online services increases the risk of children encountering content harmful to children. The risk of cumulative harm is also heightened by high usage. We therefore consider these functionalities and features to increase the likelihood of encountering all kinds of content harmful to children: -Affirmation-based functionalities such as the ability to react to content (e.g., ‘likes’) and comment on content (see 4c) can provide children with quantifiable affirmation from the engagement of other users. User connections (see 3a) can also play a role, as children may be incentivised to build a large network of connections due to the perception of popularity associated with this. This engagement or perceived popularity can be understood as social affirmation and act as a reward that causes users, including children, to use a service more often. -Alerts and notifications can result in children accessing online services more often by prompting them to check activities they were notified of or reminding them they have not visited the service recently. Alerts and notifications can also intersect with affirmation-based functionalities by signalling to children when they have received engagement from others, which can re-enforce service use. -Infinite scrolling and auto-play features can prevent interruptions or prompts to end the consumption of online content. Users can often scroll through seemingly endless content feeds where videos play automatically, making it more challenging to disengage and therefore increasing time spent on a service. These feeds are also typically curated by content recommender systems (see 6) which show content that is most likely to be engaging to that user, further increasing the likelihood that children spend more time on the service.
General risk factors all user-to-user services should take account in their children’s risk assessment
All user-to-user services
• Risk factor: User base age. The age of children on a service is one of the many demographic factors affecting the risk of harm related to content harmful to children. However, we have found evidence which suggests that children in different age groups face some distinct risks online based on the developmental stages they may be in. We expect services to consider the age-related risks posed to children and suggest that they do so in relation to the following age groups:81 -Children of 0-5 years are undergoing rapid development and have typically begun using online services with parental supervision. Children of this age can still encounter content harmful to children, particularly when unsupervised or when using the devices and accounts of family members.
General risk factors all user-to-user services should take account in their children’s risk assessment
-Children of 6-9 years are increasingly going online and becoming more independent, with parents often creating rules to manage their children’s online experiences. Evidence suggests that some children in this age group encounter content harmful to children, such as pornographic content, and incidents of exposure can have lasting impacts. -Children of 10-12 years are in a period of transition in which they often gain more independence and socialise more online, typically under less active parental supervision. This creates an increased risk of harmful interactions such as bullying content. -Children of 13-15 years are using online services more often and increased risk-taking tendencies see some actively seeking out content harmful to children such as pornographic content. The onset of mental health challenges is more common for children of this age than other age groups, heightening the impact of content harmful to children such as content promoting suicide and self-harm. -Children of 16-17 years are typically more expressive online which coincides with new rights and freedoms as well as often being unsupervised online. Given the legal and practical independence gained by children at this age, both children and parents agree that they should be granted more freedom to access content online. Evidence suggests that children in this age group have stronger media literacy competencies than younger age groups to assess and navigate more challenging content. However, children in this age group may be more likely to engage in certain behaviours that can increase their risk of encountering harmful content, such as livestreaming and creating adult profiles.
All user-to-user services
• Risk factor: Other user base demographics. The demographics of children using your service (including users’ protected characteristics, media literacy levels and mental health) will influence the risk of harm related to all kinds of content harmful to children. Overall, we have found that vulnerable child users (particularly children with multiple protected characteristics and existing mental health conditions) are more likely to experience harm from content harmful to children and are impacted differently by it. For example, we expect you to consider: -How the gender of users affects your assessment of risk – girls are more likely to encounter, and are disproportionately impacted by, various kinds of content harmful to children such as eating disorder content, abuse and hate content, bullying content, and suicide and self-harm content. Boys are more likely to encounter and engage with pornographic content as well as dangerous stunts and challenges content. -How users with protected characteristics (race, including ethnicity and minority ethnic groups, sexuality, sexual identity, religion, and disability) affects your assessment of risk, including the risk of harm to users with multiple protected characteristics. These dynamics are highly complex and context specific, and evidence is provided in the Children’s Register on user base demographics for each kind of content harmful to children. This can help you assess this risk factor even if you do not have any service-specific information on the make-up of your user base.
All user-to-user services
• Risk factor: Business model (revenue model and growth strategy). Your revenue model may increase the risk that children encounter different kinds of content harmful to children on your service. For example, we expect you to consider:
General risk factors all user-to-user services should take account in their children’s risk assessment
-How the design of your service to optimise your revenue may influence risk. For instance, to maximise engagement and therefore revenue, recommender systems on your service may suggest to children content that can be engaging but harmful (such as violent content). Or your service design may minimise ‘friction’ when sharing that content to increase engagement and therefore revenue. -How a focus on revenue maximisation could limit the effectiveness of systems and processes to protect children. For instance, services may be discouraged from adopting content moderation practices that could reduce children’s exposure to content harmful to children but also potentially reduce engagement (and in turn, revenue). -How features directly related to how a service generates revenue may increase the risk that children encounter content harmful to children. For instance, content harmful to children that is ‘boosted’82 may be promoted to children. Or paid-for advertisements83 related to content harmful to children (such as advertisements for pornographic content) may increase the risk that children subsequently search for related user-generated content on the service. Your growth strategy84 may increase the risk that children encounter different kinds of content harmful to children on your service. For instance, we would expect you to consider how a focus on growth could incentivise you to prioritise the use of your financial resources for activities aimed at growing your business (e.g., marketing campaigns, R&D activities) rather than for systems and processes to protect children. We would also expect you to consider how growing your service through implementing new technologies and offerings, particularly emerging technologies (e.g. Generative AI), of which children are often early adopters, may attract children to your service.
All user-to-user services
• Risk factor: Commercial profile.85 Your commercial profile may increase the risk that children encounter different kinds of content harmful to children. For example, we would expect you to consider: -How low-capacity or early-stage businesses may pose heightened risk if they have more limited technical skills and financial resources to introduce effective risk management. For instance, they may have insufficient resources to adopt technically advanced automated content moderation processes, or to hire a large number of paid-for moderators. -How a fast-growing user base may negatively affect risk management, given the increased scale and sophistication of the moderation technologies and processes required to keep track of content harmful to children (particularly since the sources of risk can change quickly as the user base develops).
General risk factors all user-to-user services should take account in their children’s risk assessment
* See Table 1.1 and 1.2 of the Children’s Register for more detail on how different kinds of content harmful to children have been grouped.
Kind of content
Where it is listed as a key kind of content harmful to children in the
harmful to children
Children’s User-to-User Risk Profile
1.
Pornographic content
1a. Social media services, 1e. Pornography services, 3a. User connections, 3b. Group messaging, 4b. Direct messaging, 4d. Posting images or videos, 5a. User-generated content searching, 6. Content recommender systems and 7. Functionalities and other features that increase engagement.
2.
Suicide and self-harm content
1a. Social media services, 1c. Discussion forum and chat room services, 1f. Video-sharing services, 3a. User connections, 3b. User groups, 3b. Group messaging, 4a. Livestreaming, 4c. Commenting on content, 4d. Posting images or videos, 4e. Re-posting or forwarding content, 5a. User-generated content searching, 5b. Content tagging, 6. Content recommender systems and 7. Functionalities and other features that increase engagement.
3.
Eating disorder content
1a. Social media services, 1c. Discussion forum and chat room services, 1f. Video-sharing services, 2a, User profiles, 3a. User connections, 3b. User groups, 3b. Group messaging, 4c. Commenting on content, 4d. Posting images or videos, 5a. User-generated content searching, 5b. Content tagging, 6. Content recommender systems and 7. Functionalities and other features that increase engagement.
4.
Abuse and hate content
1a. Social media services, 1d. Gaming services, 1f. Video-sharing services, 2a. User profiles, 2b. Anonymous user profiles or users without accounts, 3b. Group messaging, 4b. Direct messaging, 4c. Commenting on content, 4d. Posting images or videos, 6. Content recommender systems and 7. Functionalities and other features that increase engagement.
5.
Bullying content
1a. Social media services, 1b. Messaging services, 1d. Gaming services, 2b. Anonymous user profiles or users without accounts, 3b. Group messaging, 4b. Direct messaging, 4c. Commenting on content, 4e. Re-posting or forwarding content and 7. Functionalities and other features that increase engagement.
6.
Violent content
1a. Social media services, 1b. Messaging services, 1d. Gaming services, 1f. Video-sharing services, 2b. Anonymous user profiles or users without accounts, 3a. User connections, 3b. Group messaging, 4a. Livestreaming, 4d. Posting images or videos, 4e. Re-posting or forwarding content, 6. Content recommender systems and 7. Functionalities and other features that increase engagement.
7.
Harmful substances content
1a. Social media services, 1f. Video-sharing services, 4d. Posting images or videos, 5b. Content tagging and 7. Functionalities and other features that increase engagement.
Kind of content
Where it is listed as a key kind of content harmful to children in the
harmful to children
Children’s User-to-User Risk Profile
8.
Dangerous stunts and challenges content
1a. Social media services, 1f. Video-sharing services, 3a. User connections, 4d. Posting images or videos, 5b. Content tagging, 6. Content recommender systems and 7. Functionalities and other features that increase engagement.
9.
Content that shames or otherwise stigmatises body types or physical features
1a. Social media services, 1f. Video-sharing services, 4c. Commenting on content, 5b. Content tagging, 6. Content recommender systems and 7. Functionalities and other features that increase engagement.
10.
Content that promotes depression, hopelessness and despair
1a. Social media services, 1f. Video-sharing services, 2b. Anonymous user profiles or users without, 4d. Posting images or videos, 5b. Content tagging, 6. Content recommender systems and 7. Functionalities and other features that increase engagement.
Select Yes (Y) or No (N) for the following questions about your search service.
1. Is my service any of the following service types? a. General search service (including downstream general search service88). b. Vertical search service.
Y / N Y / N
2. Does my service have any of the following functionalities? Select all that apply: a. Provide users with search predictions or suggestions. b. Allows users to search for photographs, videos or visual images.
Y / N
Y / N
77 We have listed the key kinds of content harmful to children associated with each specific risk factor in alphabetical order.
78 We describe ‘anonymous user profiles’ as a user-to-user service functionality allowing users to create a user profile where their identity is unknown to an extent. This includes instances where a user’s identity (an individual’s formal or officially recognised identity) is unknown to other users, for example, through the use of aliases (‘pseudonymity’). It also includes where a user’s identity may be unknown to a service, for example, services that do not require users to register by creating an account. Further information on risk factors is available in Section 18 of the Children’s Register.
79 These spaces are likely to be community moderated.
80 We describe ‘bots’ as an umbrella term that refers to a software application or automated tool that has been programmed by a person to carry out a specific or predefined task without any human intervention. Further information on risk factors is available in Section 18 of the Children’s Register.
81 We recognise that service providers will likely have different levels of understanding, evidence and data about the age of users on their services. See ‘Assessing the user base’ sub-section of the Children’s Risk Assessment Guidance for more detail. However, we still expect them to consider the risk of harm to children in different age groups that they have reason to believe may be accessing the service, regardless of whether or not they intend all those age groups to do so.
82 Where users pay services to give prominence to their content. 83 For instance, display advertisements. While paid-for advertisements are not typically in scope of the Act themselves (unless they also amount to user-generated content), they are considered here by virtue of being a vector to user-generated content harmful to children.
84 We describe ‘growth strategy’ as how the service plans to expand its business, for example, through the adoption of emerging technologies. Further information on risk factors is available in Section 18 of the Children’s Register. 85 We use ‘commercial profile’ to refer to the size of the service in terms of capacity, the stage of service maturity, and the rate of growth in relation to users and/or revenue. Further information on risk factors is available in Section 18 of the Children’s Register.
Table 6. Summary of specific risk factors in the Children’s User-to-User Risk Profile associated with each kind of content harmful to children

Children’s Search Risk Profile

1.12 The Children’s Search Risk Profile is presented in Table 9. Each row represents a unique risk factor that service providers should consider when conducting their children’s risk assessment. The information provided on the risk factors is based on the evidence in Section 12 of the Children’s Register.
1.13 When consulting the table, services should do the following: • First, answer the ‘Yes’/ ‘No’ questions in Figure 2 about the characteristics of your service.86 • Second, use your answers to select which specific risk factors from Table 7 apply to you. Each ‘Yes’ answer corresponds to a risk factor you will need to account for in your children’s risk assessment. For example, if you answered ‘Yes’ to questions 1a and 2a then you should select general search services (including downstream general search services) and predictive or suggestive search as risk factors from the Children’s Search Risk Profile. A Glossary is available to help you interpret your risk factors accurately.87 • Third, review the four general risk factors at the bottom of the Children’s Search Risk Profile: user base age, other user base demographics, revenue model and commercial profile. These apply to all services, and you will need to take account of each in your children’s risk assessment.
1.14 After you have taken these three steps, you should have the list of risk factors you will need to take account of when conducting your own children’s risk assessment. This list
86 If your service offers multiple versions – for example, mobile and web – you should select ‘Y’ if any version of the service has the relevant characteristic(s). However, this only applies where versions are similar enough to be treated as a single service. 87 If, after consulting Section 18 of Children’s Register, you are still unsure if the risk factor applies to you, we would suggest you read the corresponding information provided about that risk factor in Table 9 and consider if this information is relevant to your service. You may also wish to consult Section 12 of the Children’s Register for more detailed information on the corresponding risk factor or kind of content harmful to children.
includes any specific risk factors you have selected alongside all four of the general risk factors.
1.15 Step 2 of the Children’s Risk Assessment Guidance provides details on how to use this list of risk factors as part of your children’s risk assessment. At Step 2, you will also consider how the risk factors you have selected affect your service (e.g., you have considered if this is a risk that you are already managing, or one that you may need to address through a set of mitigations to help reduce the risk of harm to children).
Figure 2. Questions for identifying your risk factors

Table 7. Children’s Search Risk Profile

General risk factors all search services should take account in their children’s risk assessment
All search services
• Risk factor: User base age.
The age of children on a service is one of the many demographic factors
affecting the risk of harm related to content harmful to children. However,
we have found evidence which suggests that children in different age
groups face some distinct risks online based on the developmental stages
they may be in. We expect services to consider the age-related risks posed
to children and suggest that they do so in relation to the following age
groups:89
-Children of 0-5 years are undergoing rapid development and have
typically begun using online services with parental supervision.
Children of this age can still encounter content harmful to children,
particularly when unsupervised or when using the devices and
accounts of family members.
-Children of 6-9 years are increasingly going online and becoming
more independent, with parents often creating rules to manage
their children’s online experiences. Evidence suggests that some
children in this age group encounter content harmful to children,
such as pornographic content, and incidents of exposure can have
lasting impacts.
-Children of 10-12 years are in a period of transition in which they
often gain more independence and socialise more online, typically
under less active parental supervision. This creates an increased
risk of encountering harmful content online.
-Children of 13-15 years are using online services more often and
increased risk-taking tendencies see some actively seeking out
content harmful to children such as pornographic content. The
onset of mental health challenges is more common for children of
this age than other age groups, heightening the impact of content
harmful to children such as content promoting suicide and self-
harm.
-Children of 16-17 years are typically more expressive online which
coincides with new rights and freedoms as well as often being
unsupervised online. Given the legal and practical independence
gained by children at this age, both children and parents agree that
they should be granted more freedom to access content online.
Evidence suggests that children in this age group have stronger
media literacy competencies than younger age groups to assess
and navigate more challenging content. However, children in this
age group may be more likely to engage in certain behaviours that
can increase their risk of encountering harmful content.
All search services
• Risk factor: Other user base demographics. The demographics of children using your service (including users’ protected characteristics, media literacy levels and mental health) may influence the risk of harm related to all kinds of content harmful to children. Overall, we have found that vulnerable child users (particularly children with multiple protected characteristics and existing mental health conditions) are more likely to experience harm from content harmful to children and are impacted differently by it. We would expect you to consider these dynamics when you assess the risk of each type of content harmful to children. These dynamics are highly complex and context specific, and evidence is provided in the Children’s Register on user base demographics for each kind of content harmful to children. This can help you assess this risk factor even if you do not have any service-specific information on the make-up of your user base.
All search services
• Risk factor: Business model (revenue model and growth strategy). We would expect you to consider if and how your revenue model may increase the risk that children encounter different kinds of content harmful to children on your service.90 Your growth strategy may also increase risk. For instance, we would expect you to consider how the development of an expansive search index may
influence risk, if not accompanied by appropriate systems and processes that ensure indexed webpages containing content harmful to children are not returned in the search results for child users.
All search services
• Risk factor: Commercial profile.
Your commercial profile may affect the likelihood that children encounter
content harmful to children. For example, we expect you to consider how
low-capacity services91 and early-stage services92 may have limited
capacity/resources to develop effective processes (such as technically
advanced moderation processes) to minimise the likelihood of children
encountering content harmful to children.
88 For the purposes of the Children’s Search Risk Profile and our recommended measures, a downstream general search service is still a ‘general search service’. The unique business models of downstream general search services do not change the type of search service that is offered to users. That being said, it will be relevant when determining who the ‘provider’ of the downstream general search service is. We consider it is for the entities involved in the downstream arrangement to determine how many distinct search services are being offered to users, and who the ‘provider’ is for each of these services. If entities involved in a downstream general search service arrangement fail to agree on who the ‘provider’ of the service is and make arrangements to ensure compliance with the Act, each risks being the entity that we determine to be the ‘provider’.
89 We recognise that service providers will likely have different levels of understanding, evidence and data about the age of their users. See ‘Assessing the user base’ sub-section of the Children’s Risk Assessment Guidance for more detail. However, we still expect services to consider the risk of harm to children in different age groups that they have reason to believe may be accessing the service, regardless of whether or not they intend all those age groups to do so.
90 While general search services typically generate revenue using an advertising-based model and there is a risk that advertising could be used to promote content harmful to children, there is very limited evidence demonstrating any direct link between different revenue models and the presence of content that is harmful to children in search results. We nevertheless encourage search services to consider any links between their revenue model and increased risks for children.
91 In terms of number of employees or revenues. See Section 18 of the Children’s Register. 92 Search services that are start-ups or at an early growth stage. See Section 18 of the Children’s Register.

Section 2 Non-designated content

In this sub-section we provide guidance to service providers to help them fulfil their duties to assess the risk presented to children by non-designated content harmful to children, and (for user-to-user services) to notify Ofcom where a children’s risk assessment identifies presence of non-designated content.

What is non-designated content?

2.1 Non-designated content is set out in the Act as content “of a kind which presents a material risk of significant harm to an appreciable number of children in the United Kingdom”.93 There are some exceptions to what can be considered to be non-designated content.94 All service providers completing a children’s risk assessment have a duty to assess the risk of children encountering non-designated content on their service.

Identify non-designated content for assessment

2.2 Ofcom has identified two categories of non-designated content which you should consider as part of your children’s risk assessment.95 When considering the risk of harm to children from encountering these categories of non-designated content you should take into account Ofcom’s Children’s Risk Profiles.
2.3 You should also consider whether there are other kinds of non-designated content that Ofcom has not identified in the Children’s Register that you should include in your assessment.96 To do this, you should consider Ofcom’s three-step framework for identifying non-designated content. This is set out in the Introduction to the Children’s Register and is based on the terms set out in the statutory definition above. Ofcom interprets these terms as follows: a) ‘Significant harm’ means physical or psychological harm that causes lasting impact on children’s wellbeing, attitudes or behaviour, whether directly to users or indirectly to others; b) ‘Material risk’ relates to the likelihood of harm occurring, and this can be assessed by linking significant harm to a specific kind of content; and c) ‘Appreciable number’ does not necessarily have to be a very large number of children, but not one child or very few children.
2.4 If you identify kinds of content that you believe meet the definition of non-designated content, you should assess the risk of harm to children presented by each of these kinds, as set out in the sub-section ‘Assess non-designated content’ below.
93 Section 60(2)(c) of the Act.
94 Content is not to be regarded as non-designated content where the risk of harm flows from the content’s potential financial impact; the safety or quality of goods featured in the content; or the way in which a service featured in the content may be performed. 95 These are: Content that shames or otherwise stigmatises body types or physical features (‘body stigma content’); and Content that promotes depression, hopelessness and despair (‘depression content’). 96 For example, if you are aware of a new or emerging kind of content on a service that is similar to the one you provide.
2.5 You must make a record of any kind(s) of non-designated content you identify for assessment on your service. You must also record how you have considered the risk of children encountering non-designated content by means of your service, including when you do not identify any kind(s) of non-designated content for assessment, beyond those identified by Ofcom.

Assess non-designated content

2.6 You should consider likelihood and impact to assign a level of risk to any kind(s) of non-designated content you have identified for assessment. This includes the two kinds of non-designated content that Ofcom has identified, and any additional kind(s) of non-designated content you have identified for assessment.
2.7 This means you should have at least two risk levels for non-designated content as an outcome of Step 2, and additional risk levels for any other kind(s) of non-designated content you have identified for assessment. This is important when considering which safety measures you should implement as part of Step 3.
2.8 When assigning a level of risk to any kind(s) of non-designated content you have identified for assessment, you should also consider evidence to understand the incidence of such kind(s) on your service.
2.9 You should make a record of the level of risk assigned to each of the two kinds of non-designated content that Ofcom has identified, and to any additional kind(s) of non-designated content you identified for assessment.

Notify Ofcom of non-designated content

2.10 A user-to-user service provider that identifies the presence97 of non-designated content on their service in their children’s risk assessment has a duty to notify Ofcom of:98 a) The kinds of such content identified, and b) The incidence of those kinds of content on your service.
2.11 We consider that incidence in this context refers to the amount of that relevant non-designated content on your service. When considering the incidence of that kind of content, you should refer to appropriate core and/or enhanced evidence inputs, which will vary depending on the service (e.g., user complaints and reports, evidence drawn from existing controls, results of content moderation systems or evidence from external experts).
2.12 You should notify Ofcom using the following email address: nondesignatedcontent@ofcom.org.uk.
97 This means where you have assessed the level of risk from a kind of non-designated content as low, medium or high, but not negligible. 98 Section 11(5) of the Act.

Section 3 Evidence inputs

This sub-section of the guidance focuses on the different types of evidence that service providers should consider when assessing risk of harm to children.

Why is evidence important?

3.1 The purpose of the children’s risk assessment is to improve your understanding of how harm to children could take place on your service, and what safety measures you need to put in place to protect children. The guiding principle when deciding what evidence to collect should be what will enable you to make an accurate children’s risk assessment that meets the suitable and sufficient standard.
3.2 To be suitable and sufficient, your children’s risk assessment must include all the elements of a children’s risk assessment specified in the Act (section 11(6) for user-to-user services and section 28(5) for search services). It should be specific to your service and reflect the risks accurately.
3.3 It is important that you understand the risk of harm to children presented by each kind of primary priority content, each kind of priority content, and by non-designated content, as well as how the design and use of your service affects the level of risk of harm to children, to implement appropriate safety measures. This means that your judgements on risk should be based on relevant information and evidence.

How should you decide what evidence is relevant?

3.4 We understand that the appropriate level of evidence will vary based on the size and nature of the service. • We expect all service providers to use core evidence inputs as part of their assessment. We provide advice on core types of evidence that all service providers should consider when assessing their level of risk for a particular kind of content harmful to children. • You should use enhanced types of evidence if your service is large, if you have identified several specific risk factors, or if you are not confident that the core inputs have enabled you to make an accurate assessment of the risk of children encountering content harmful to children on your service.
3.5 When considering user data, we encourage you to familiarise yourself with the data protection legislation, and to consult the ICO’s guidance on UK GDPR requirements and the Children’s Code.99
3.6 In Step 2 (assess the risks), you should review your evidence to assess the likelihood and impact of children encountering each particular kind of content harmful to children.
3.7 In this step, you should consider for each kind of primary priority content, each kind of priority content, and for any non-designated content you have identified for assessment, all the core inputs and any other relevant information you already hold. Then, at each stage of
99 ICO, UK GDPR guidance and resources; ICO, Age appropriate design: a code of practice for online services; ICO, ‘Likely to be accessed’ by children guidance.
your analysis, you should consider whether you have sufficient information to reach accurate conclusions on the level of risk for that particular kind of content. If not, you should consider gathering additional evidence from the list of enhanced inputs.
3.8 Each kind of input is explained in the next sub-sections. While we expect the core and enhanced inputs to provide service providers with a good understanding of risk, the suggestions are not exhaustive. You should consider whether you need to use any further evidence.

Evidence on child age

3.9 For the purposes of the children’s risk assessment, we expect you to use the best available information about the users on your service. When considering user data, we encourage you to familiarise yourself with the data protection legislation, and to consult the ICO’s guidance on UK GDPR requirements and the Children’s Code.100
3.10 Different kinds of evidence can give insights about the age of individual users, or the age distribution of the overall user base with varying degrees of accuracy. While this may vary by service provider, some examples of relevant information could include: a) Data from age assurance processes; b) Information gathered by the service on the users’ age in the process of setting up an account c) Proxy indicators for age, such as behavioural patterns identified while a user is active on the service which gives a reasonable indication of age; d) Consumer or other research conducted to inform product or advertising decisions, or to make the service attractive to children; e) Evidence about children’s habits online that may indicate whether your service is likely to attract users of a certain age; and f) Any other third-party research or research about the age(s) of the user base on your service.
3.11 You should not rely on self-declared age as an accurate indication of the age of users as children sometimes provide inaccurate self-declaration information.101 Relying on this as evidence could significantly underestimate the total number of children on your service overall, or within particular age groups.
3.12 Similarly, you should not rely on any minimum age requirement in your terms of service as evidence of the actual age of children on the service. Unless you are using highly effective age assurance to enforce your minimum age, it is likely there will be some underage users accessing the service and you should take a conservative view of the potential number and age of children for the purposes of the children’s risk assessment, erring on the side of caution when assessing the scope for children (including children below your minimum age requirement) to be impacted by harmful content.

What is a core input?

3.13 We expect you to use core evidence inputs to complete your children’s risk assessment.
100 ICO, UK GDPR guidance and resources; ICO, Age appropriate design: a code of practice for online services; ICO, ‘Likely to be accessed’ by children guidance. 101 See Section 17 of the Children’s Register: ‘Age Groups’.
3.14 Some service providers may be able to rely on core inputs alone to confidently assess risk of harm to children. This could be the case where: • You identify no or very few risk factors from Ofcom’s Children’s Risk Profiles relating to content harmful to children (suggesting a potential lower level of risk on the service); and • Evidence from the core inputs enables you to determine with confidence the likelihood and impact of children encountering content harmful to children on your service and to use the Risk Level Table to assign a risk level to this kind of content.
3.15 You should also consider any information you already hold, including past children’s risk assessments, that may be relevant to your assessment. Table 8 includes examples of core inputs.
3.16 Failing to consider all core inputs may mean that your children’s risk assessment is not suitable and sufficient.

Table 8: Core evidence inputs

Core inputs
Explanation
Children’s Risk Profiles, relevant parts of Ofcom’s Children’s Register of Risks and Guidance on Content Harmful to Children
You have a duty to take account of the relevant Children’s Risk Profile as part of your risk assessment. We encourage service providers to consult the relevant sub-section of the Children’s Register, to understand the context of the risk factors in the Children’s Risk Profiles. You can prioritise the most relevant parts of the Children’s Register once you have established your risk factors in Step 1. The Children’s Register looks at each kind of content harmful to children, in some cases grouping similar kinds of harmful content. Each section presents a summary box and full analysis of risk factors associated to one or more particular kinds of content harmful to children.
User complaints, including user reports
Under the Act, you are required to provide complaints procedures which are easy-to-access and use. These should allow complaints and reports to be made by children and for you to take appropriate action. You should consider any data from these complaints and reports when carrying out your children’s risk assessment. Services should try to ensure they have a means to collect and interpret complaints and reports from child users. If you have not collected this information before and set up a new user reporting function, you should consider any reports when you update your children’s risk assessment. Evidence gathered from this input could include, for example, the kind of content harmful to children being complained about, any relevant data on the children making these complaints if this is already available to you, the accuracy of complaints, and the length of time taken for an appropriate action to be taken. This input should help you understand the impact and frequency of a particular kind of content harmful to children on your service. User complaints will help you assess the likelihood (how many children’s
complaints) or impact (the nature of children’s complaints) of children encountering content harmful to children on your service.
Where relevant, user data including age insights
By user data we mean data you hold that has been provided by users, including their personal data (e.g., data provided when a user sets up an account), and data about users that you have created, compiled or obtained (e.g., data relating to when or where users access a service or how they use it). You may already hold this kind of user data, for example, for analysis via behaviour identification technology or user profiling technology. Service providers have a duty to consider the age of children on their service in relation to their assessment of each particular kind of content harmful to children, and can do so with any relevant evidence they hold on user age. Relying only on self-declared age information would not be considered sufficient evidence to draw firm conclusions about the age of children on a service. We would expect service providers to assume that the true number of children using their service – both overall and within specific age groups – could be significantly higher than suggested by self-declared information (or similar forms not considered robust). Considering user data, in combination with other inputs into the children’s risk assessment, will help you understand if any particular child user groups are at risk of encountering content harmful to children on your service. In particular, you should consider reports or findings relevant to children’s experiences online or on your service. This is relevant because, as set out in the Children’s Risk Profiles and Children’s Register, certain content can disproportionately affect certain demographic groups (e.g., children with mental health conditions are more likely to encounter content promoting suicide, self-harm and eating disorders). User data will therefore help you determine the impact of each particular kind of content harmful to children on your service. User data also includes any data held as a result of age estimation and age verification processes. When considering user data, you must also consider privacy rights and your duties under the UK General Data Protection Regulation (GDPR). This is likely to be of greater consideration for age assurance and age verification measures. We encourage you to consult the ICO’s guidance.102
Retrospective analysis of incidents of harm to children
Following any significant incident of harm to children experienced on your service, you should undertake retrospective analysis or a ‘lessons learned’ exercise. This information should inform your children’s risk assessment. A significant incident could include, for example, a major incident that causes serious harm, a prominent trend in content
harmful to children, or an individual piece of content which becomes widely disseminated. Retrospective analyses will help you assess the impact of content harmful to children on your service, particularly in relation to content which is less common but high impact, and to obtain any age-related insights, for example, more severe impacts for younger children. Such case studies may allow service providers to examine how particular aspects of the service’s design (such as user characteristics, functionalities, recommender systems) may have played a role and where mitigating measures (such as content moderation, terms of service, user reporting) and associated processes could have been more effective.
The findings of your illegal content risk assessment
All service providers completing a children’s risk assessment will also have to complete an illegal content risk assessment. The findings of the illegal content risk assessment should include assessment of the features, functionalities, user base and characteristics of the service. Elements of this assessment will be relevant to your children’s risk assessment as some illegal content may be related to, or overlap with, some content harmful to children. For example, content that encourages or assists suicide, in addition to being priority illegal content, also amounts to primary priority content (content which encourages, promotes or provides instructions for suicide) for the purposes of the children’s risk assessment.
The findings of
The findings of your children’s access assessment may help you to understand your user base.
your children’s
access
assessment
Evidence drawn from existing controls
If you already have existing controls to mitigate risk as part of the design or operation of your service, then you may want to consider how these controls impact the level of risk posed to users on your service. You should use evidence drawn from these processes which demonstrates their effectiveness in mitigating a particular kind of content harmful to children. For example, this may include results of content moderation systems. Most service providers are likely to have a content moderation system in place on their service, though the scope and maturity of these systems varies significantly between services. Assessing the effectiveness of content moderation decisions and the systems themselves also helps you to understand the level of mitigation provided by this measure in your children’s risk assessment. This may also include results from age assurance processes where a service has an existing means of determining the age or age range of a particular user. Including this in your children’s risk assessment will help you understand the likelihood of children encountering content harmful to children on your service, the effectiveness of your mitigation measures,
and the effect of different characteristics of your service on risk levels (e.g., if a product change increases/reduces the amount of content harmful to children you detect).
Other relevant information
Consider if you already have the enhanced inputs in Table 9 (e.g., the results of testing or any research commissioned) to support your children’s risk assessment. This may include any existing reporting on content, user engagement and complaints (particularly negative signals from children regarding suggested content) published research, referrals you have made to law enforcement, reports provided to you by expert groups or by law enforcement agencies, data on user behaviour relating to content harmful to children, or the outcomes of product testing. This could also include considering the findings of other assessments or audits you have completed as part of other regimes or internal processes. In addition, service providers may have evidence or data about specific features or functionalities. For example, from product testing them, optimising the design of the service, or running A/B tests to understand their adoption, including by child users. We would expect the availability of this information to be considered when risk assessing service features and functionalities, including to inform how the service might be used by children. Any other information you hold that can support your risk assessment.
Illustration: Using core and enhanced evidence inputs
Using the Children’s Risk Profiles, service provider A has identified some risk factors
suggesting its service may present an increased risk of harm to children for a particular
kind of content harmful to children. Service provider A gathers evidence from all the core
inputs, but it finds very little to no evidence of this kind of content, despite the Children’s
Risk Profile indicating otherwise. There is therefore some ambiguity about the risk of harm
to children from encountering this content, and the service provider is unable to
determine accurately its likelihood or impact, and unable to assign a risk level using the
guidance in the Risk Level Table. Service provider A then consults the list of enhanced
inputs and selects inputs which will give more evidence relating to its risk factors to help it
assign an accurate risk level and put appropriate measures in place.
Using the Children’s Risk Profiles, service provider B, has identified some risk factors
suggesting its service may present an increased risk of harm to children from a particular
kind of content harmful to children. Service provider B gathers evidence from all the core
inputs and finds evidence of this kind of content. This means the service provider has
sufficient evidence to assess the likelihood and impact of children encountering this kind
of content on its service, and therefore it can assign itself a risk level for this kind of
content using the Risk Level Table, and without including additional evidence from the
enhanced inputs.
Using the Children’s Risk Profiles, service provider C has identified no specific risk factors
which suggests it does not present an increased risk of harm to children for a particular
kind of content harmful to children. To assess whether this is correct, service provider C
gathers evidence from all the core inputs and finds multiple sources of relevant evidence
which shows it is not at high risk for this kind of content. This means the service provider
can determine the likelihood and impact of children encountering this kind of content on
its service and assign a risk level for this kind of content using the Risk Level Table, and
without including additional evidence from the enhanced inputs.
Results of product testing103
To improve your understanding of risks to children on your service at a product level, you may consider running tests on individual products ahead of launching them on your wider services to understand how users behave and engage with the products and the potential impact of any behavioural biases. Evaluating data and insights gathered from these tests may improve your children’s risk assessment, because testing can indicate the potential effect of any product changes and help determine whether the effect may increase or decrease the risk of harm to children from encountering content harmful to children on your service. For example, services running on-platform tests104 of their recommender systems should include any additional safety metrics (such as prevalence) they gather as part of this routine testing to provide insights as to how frequent design adjustments may impact the risk of content harmful to children being disseminated on the service. Considering the results of product testing will help you understand certain risk factors which you may have identified in the Children’s Risk Profiles, such as functionalities which allow child users to find and encounter content, to communicate with one another, and/or to network. This kind of input is particularly relevant for services carrying out a new risk assessment relating to a proposed significant change. Product testing is particularly likely to be appropriate where you are launching new features that have not been widely deployed before, or if features are specific to child users.
Consultation with internal experts on risks and safety measures
To improve your understanding of a specific risk child users may face, or a technical measure to mitigate such a risk, you should consult with experts with specific knowledge and expertise on child protection as well as child rights. A thorough examination process for a technical safety measure should consist of regular thematic technical expert meetings supported with focused follow up work. This examination process should provide a clear understanding of how technical measures, systems and processes may help address risk. Consultation should happen regularly, and records of the engagement should feed into an annual risk assessment review, or experts can be brought into the four-step process while the children’s risk assessment is underway to provide formal and targeted input.
Views of independent experts
This input is likely to be valuable when considering complex topics such as the intersection of different kinds of content harmful to children, mitigation measures and the rights of children, including their freedom of expression. Consulting with child safety and child development experts can help you consider how content harmful to children manifests online in general and/or on your service specifically, which would in turn help you develop mitigation and management techniques which are targeted and effective. You should take steps to ensure the quality and accuracy of any third-party advice. Other types of expert consultation may also be relevant for your service to consider. This could include a view of experts on industry trends, regulatory standards and the views of certain trade bodies or technical experts in relevant fields.
Internal and external commissioned research
If you are seeking to access additional expert resource and expertise to incorporate into your children’s risk assessment, you may commission internal and/or external research. For instance, research could focus on specific trends or content, informing service providers’ approach to safety and moderation on the service relevant for child safety. The purpose of this input is that expert research would allow your service to improve its understanding of the factors which may drive content harmful to children appearing on your service, child-specific behaviours and developmental considerations, the impact of that content, and how it may be mitigated effectively.
Outcomes of external audit or other risk assurance processes
To improve your confidence that your trust and safety processes or wider risk management systems are comprehensive to protect children, you may commission a third party to audit aspects of your service or undergo another form of risk assurance process. Independent audits can provide insights and analysis which services are unable to produce or assure themselves. They offer services the chance to be assessed and to identify new ways of improving their trust and safety processes. Services and any third-party suppliers should take steps to ensure that any methodology applied is well thought out and that the assurance process provides an independent and objective assessment of performance and recommendations for improvement, and that assessors understand your child users. Including the outcomes of these audits in the children’s risk assessment process can provide greater independence and granularity of detail as to the accuracy and quality of the children’s risk assessment. Services that lack in-house capacity to carry out these processes may benefit from seeking third-party audits; some services may also choose to work with third parties to seek independent and objective scrutiny of their risk assessment processes.
Consultation with users and user research
To improve your understanding of child user experience or the experience of a specific group of users on your service, you may engage in consultation with users (including children) or carry out other forms of user research with children. You can choose the method and frequency of consultation and how you wish to undertake this engagement with users – this could include a platform-wide initiative which gives children an opportunity to give feedback on aspects of the service, or more targeted consultation with a specific group on specific issues which the platform has reason to believe will impact them. Alternatively, you may wish to contract external agencies to deliver qualitative research or other studies, and obtain objective user feedback. This input will help your service embed safety by design. The research should complement existing user design processes but maintain a focus on understanding how children might interact with a new product or service. Research could focus on what behavioural factors (such as behavioural biases or child development) could be present at primary points in a child user’s journey that might impact on their decision-making and increase the risk of them encountering content harmful to children. This could be particularly important if a product or service is intended to operate as part of a broader ecosystem rather than on a stand-alone basis. A continued dialogue with child users of your service will help to ensure that safety features, mitigations, and other primary points of engagement (for instance, terms of service) are accessible and meet the needs of child users. Engagement could be general or designed to target specific users, such as children with vulnerabilities or in certain age groups.
Engaging with relevant representative groups
You may choose to engage with relevant representative groups to improve your understanding of the risk of harm to children. To do this, your service may reach out to organisations representing specific groups to help give these groups a channel through which they can directly feedback any concerns around the handling of content harmful to children on your service. This is a helpful action to take if your service has evidence that certain groups of children, such as those in specific age groups, will be particularly impacted by an aspect of your service’s design, and particularly beneficial for services reviewing their children’s risk assessment following a significant change to an aspect of its service design.
102 ICO, UK GDPR guidance and resources; ICO, Age appropriate design: a code of practice for online services; ICO, ‘Likely to be accessed’ by children guidance.

What is an enhanced input?

3.17 We expect that you should also use enhanced inputs if your service is large, you have identified several specific risk factors, or if you are not confident that the core inputs have given you an accurate assessment of the risk of harm to children from encountering a particular kind of content harmful to children to allow you to undertake a suitable and sufficient assessment. This is likely to be the case where: • Evidence from the core inputs does not enable you to determine the likelihood and impact of children encountering a particular kind of content harmful to children on your service, and therefore you are unable to assign a risk level for each of the four kinds of primary priority content, each of the eight kinds of priority content, and for any kind(s) of non-designated content you have identified for assessment; • You identify several risk factors for a particular kind of content harmful to children which means you are more likely to need to consider enhanced inputs relating to that content on your service; or • You are a larger service with the resources to undertake a more thorough assessment by including enhanced inputs. This is likely to materially improve the quality of your children’s risk assessment. We would generally expect large services (those with more than 7 million UK users) to use more than the core inputs.
3.18 Large service providers or service providers that have identified several specific risk factors on their service for a particular kind of content harmful to children using the Children’s Risk
Profiles will typically need to include some or many enhanced inputs to ensure their children’s risk assessments are suitable and sufficient.
3.19 Table 7 illustrates some examples for how to decide on evidence inputs for providers of small or medium services.
Box 3: Illustrative examples of how to decide on evidence inputs
3.20 The type and number of enhanced inputs to consider when assessing the risk of a particular kind of content harmful to children is down to you. This decision is likely to be informed by the risk factors you have identified in the Children’s Risk Profiles and the size of your service.
3.21 In general, you should use as many of the enhanced inputs as you need to provide you with a clear and detailed understanding of the risks to children your service poses. We provide descriptions of the different types of enhanced inputs in Table 9 to help you decide if an input is relevant to your assessment of risk to children.
Table 9: Enhanced evidence inputs
Enhanced inputs Explanation
103 When we use the word ‘product’ we are using it as an all-encompassing term that includes any functionality, feature, tool or policy that you provide to users for them to interact with through your service. This includes but is not limited to terms and conditions, content feeds, react buttons or privacy settings. By ‘testing’ we mean services should be considering any potential risks of technical and design choices, and testing the components used as part of their products before the final product is developed. We recognise that services, depending on their size, could have different employees responsible for different products and that these products are designed separately from one another.
104 By ‘on-platform testing of recommender algorithms’ we mean the process of testing two or more variants of recommender system before proceeding with the design change. This could include but is not limited to A/B/x Testing or Multi Arm Bandit (‘MAB’) Testing.

Section 4 Risk Level Table for content harmful to children

4.1 This sub-section presents a Risk Level Table (Table 10) to use as part of Step 2 in the children’s risk assessment process. Using this table should help you to classify the risk level into high risk, medium risk, low risk and negligible or no risk, for each of the four kinds of primary priority content, each of the eight kinds of priority content, and for any kind(s) of non-designated content that you have identified for assessment. Calibrating risk levels should be based on all of the evidence you have gathered to conduct your children’s risk assessment.
4.2 The table captures considerations relevant to impact and likelihood separately. If you provide a user-to-user service, you should consider the number of risk factors you have identified in the Children’s Risk Profiles that are likely to apply to your service.105 If you provide a search service, instead of the number of risk factors, you should rely on information in the Children’s Register and your own evidence to assess likelihood.
4.3 Likelihood and impact may often be correlated. If your assessment of these diverges (e.g., relatively low likelihood and relatively high impact, or vice versa), you will need to reach a view on the risk level overall, taking account of the guidance in the table. Where the evidence is not conclusive regarding the appropriate level of risk, we expect providers to err on the side of caution and select the higher risk level, being mindful of the harm that may result to children from underestimating the level of risk.
4.4 In addition to these tables, we also provide illustrative examples in Appendix A of different risk levels for hypothetical services, to help you calibrate your risk level appropriately.
4.5 You should not apply the Risk Level Table mechanically. It is intended to help inform your assessment of a risk level, rather than to determine risk levels. However, if you have evidence of a substantial amount of content harmful to children on your service, then we would generally expect you to assess as high risk for that kind of content.

Table 10: Risk Level Table

Risk
Description
Your service may decide on this risk level if some of the
level
following conditions are met
High risk
You assess that
Any of the following applies:
there is a high
• There is significant evidence of this kind of content harmful to
likelihood that a
children106 being encountered on your service (e.g., from
child could
complaints) or it being very likely to be encountered on your
encounter this kind
service (e.g., evidence from external experts).
of content on your service
• Or your service does not prohibit this kind of content* and there
are no effective systems and processes in place to address this
content nor other factors which reduce risks to children.
• Or you have identified many107 specific risk factors in the
relevant Children’s Risk Profiles** (used in Step 1) which
increase the likelihood of this kind of content occurring and
there are no effective systems and processes in place to address
this harm nor other factors which reduce risks to children.
You assess that there is high impact to children using your service or other children from this kind of content
Any of the following applies: • Even if you have controls/existing measures in place for this kind of content, there is evidence of (1) a substantial amount of the content being present on your service; or (2) actual harm to many child users or other children.108 Evidence of either (1) or (2) is a strong indicator of being high risk and we would normally expect all services with such evidence to be high risk for that kind of content. • Based on your analysis of core and enhanced evidence inputs, you identify that this kind of content is known to have a severe impact for child users or other children, even if the number of children affected is low. For example, this could be because of the nature of the content encountered by children, the way in which it is encountered, or because the affected children are part of a vulnerable group. • There is broad scope for this kind of content to impact many children, either child users or other children. This is more likely if one or more of the following applies: ο Your service has over 1 million monthly active UK users (‘monthly UK users’),109 who are children; ο Or your service is likely to have a very high frequency of use among children or very high time spent by children; ο Or your service actively targets children as a primary audience, with large volumes of content or many kinds of content and functionalities that are strongly appealing to children;110
ο Or content can be shared and disseminated on your service in a way that has the potential to become viral rapidly and affect many children. Having the potential to impact a large number of children is not necessarily – by itself – a strong indicator of being high risk, especially compared to substantial evidence of content being present or having a severe impact on children or children in vulnerable groups. However, a service with a number of children that meets or exceeds the indicative values provided above would need robust evidence to demonstrate it does not pose high or medium risk. You should also exercise caution when interpreting evidence about age of users, as discussed in Step 2.
Medium risk
You assess that there is a moderate likelihood that a child could encounter this kind of content on your service
Any of the following applies:
• There is evidence of this kind of content being (or likely to be)
encountered on your service.
• Or you have identified several111 specific risk factors in the
relevant Children’s Risk Profiles* (Step 1) which increases the
likelihood of children encountering this kind of content and,
while you may have some systems and processes in place to
address this content, you cannot demonstrate they are effective
at reducing risks to child users or other children, nor are there
any existing controls and/or aspects of your service’s design or
operation that sufficiently reduces risks.
• Or your service does not prohibit this kind of content** and
there are some systems and processes in place to address this
content but you cannot demonstrate they are effective at
reducing risks to child users or other children.
You assess that there is moderate impact to children using your service or other children from this kind of content
Any of the following applies: • Even if you have controls/existing measures in place for this kind of content, there is evidence of (1) a moderate amount of content being present on your service; or (2) actual harm to a moderate number of child users or other children. Evidence of either (1) or (2) is a strong indicator that your service is at least medium risk, and we would normally expect all services with such evidence to be medium or high risk for that kind of content. Such evidence could be from the core or enhanced evidence inputs. • Or based on your analysis of evidence, you identify that this kind of content would have a moderate impact for child users or other children. • Or there is some scope for this kind of content to impact many children, either child users or other children. This is more likely if one or more of the following applies: ο Your service has between 100,000 and 1 million monthly UK users who are children; ο Or your service is likely to have high frequency of use among children or high time spent by children;
ο Or your service includes content or functionalities that are strongly appealing to children; ο Or content can be shared and disseminated on your service in a way that has the potential to affect several children. Having the potential to impact many users is not necessarily – by itself – a strong indicator of being medium risk, especially compared to evidence of a moderate amount of a kind of content.
Low risk
You assess that there is a low likelihood that a child would encounter this content on your service
Any of the following applies:
• You have used evidence to assess whether children could
encounter this kind of content by means of your service and
have concluded that there is no evidence of this, and you have
identified no or few specific risk factors associated with the kind
of content in the relevant Children’s Risk Profiles.*
• Or there are comprehensive systems and processes in place, or
other factors which sufficiently reduce risks of children
encountering this kind of content and your evidence shows they
are very effective (e.g., all content is pre-moderated). Even if
taking all relevant measures in Ofcom’s Protection of Children
Codes, this may not always be sufficient to mean your service is
low risk.
You assess that there is limited impact to child users or other children from this kind of content
Any of the following applies: • Even if there are occasional examples of the kind of content occurring on your service, there is limited potential for this kind of content to impact child users or other children. For example, this could be because the content impacts very few child users or other children and the severity of that content is low. If the severity of the content were high, then even a small number of instances would mean your service was not low risk.
Negligibl e or no risk
If your evidence shows it is not possible or extremely unlikely that children could
encounter this kind of content by means of your service, you may assess the risk of this
content as ‘negligible or no risk’, irrespective of whether you meet one or more of the
conditions for low/medium/high risk as defined in the rows above. If you have some of the
relevant risk factors and it is possible for children to encounter this kind of content on your
service, then to assess as ‘negligible or no risk’ for that kind of content, you will normally
need comprehensive evidence to demonstrate that your service does not pose low,
medium or high risk.
* For avoidance of doubt, a service that prohibits a kind of content in question may still have a high or
medium risk for that kind of content, depending on the assessment of relevant evidence and factors as
set out in this guidance.
** For the avoidance of doubt, a service with few risk factors for the kind of content in question may
still have a high or medium risk for that kind of content, depending on the assessment of relevant
evidence and factors as set out in this guidance.
Additionally, when assessing a kind of non-designated content that is not identified in Ofcom’s
Children’s Register and Children’s Risk Profiles, you should form your own assessment of what the
relevant risk factors are likely to be and how many apply to your service.
105 When assessing a kind of non-designated content that is not identified in Ofcom’s Children’s Register of Risks and Children’s Risk Profiles, you should form your own assessment of what the relevant risk factors are likely to be and how many apply to your service. 106 Hereafter ‘content’, referring to each kind of primary priority content, each kind of priority content, and any kinds of non-designated content you have identified for assessment.
107 We consider ‘many’ to be a large number of risk factors in proportion to the total number of specific risk factors for a particular kind of content harmful to children in the Children’s Risk Profiles. The number of risk factors we have identified for different kinds of content harmful to children in the Children’s Risk Profiles varies in line with the evidence available and the way harm manifests. A kind of content harmful to children which involves many different offences, pathways and behaviours may have more evidence available, and in turn more risk factors associated with it than one which has fewer. Therefore, for a kind of content harmful to children with a small total number of risk factors, even a few may be considered ‘many’. Similar considerations apply to ‘several’ and ‘few’ for the risk levels in the rest of the table. 108 For example, children who are indirectly affected. One example could be violent content, where links have been found between this kind of content and specific acts of violence. See Section 7 of the Children’s Register.
109 As calculated in accordance with the methodology set out in the Protection of Children Codes of Practice from paragraph [XX]. 110 This condition sets a substantially higher bar than the bar set to describe a service “of a kind likely to attract a significant number of users who are children” in the Children’s Access Assessment Guidance (i.e., a service meeting the child user condition in Stage 2 of the assessment). In the latter case, a service may meet the child user condition even if it does not actively target children, for several reasons (e.g., the service has the potential to benefit children).
111 This is intended as an overall guide, but rather than focusing purely on the number of risk factors, you should consider the combined effect of the risk factors to make an overall judgement about the level of risk on your service.
Source: Ofcom

Section 5 Making a significant change to your service

5.1 This sub-section of the guidance explains the circumstances which could amount to a significant change to your service where you may need to carry out a new children’s risk assessment relating to that proposed change.

Carry out a new risk assessment before making a significant change to your service

5.2 If you plan to make a significant change to your service, you must carry out a new children’s risk assessment before making the change. This applies to a change to any aspect of the service design or operation which is reasonably likely to have a significant impact on the risk of harm to children. This is a specific legal duty, so you should carefully consider proposed changes to your service in advance of putting them into operation.
5.3 The legal duty requires you to carry out a new children’s risk assessment as it relates to the impacts of the proposed change. In practice, this may require you to carry out a new assessment of the whole service. To ensure that your children’s risk assessment is suitable and sufficient, you should consider whether the impact of the change may affect other parts of the service and undertake a children’s risk assessment accordingly. This new assessment can be done by carrying out each step of the four-step process.
5.4 This duty is intended to capture changes which could impact the risk of harm to children presented by content harmful to children and the design and use of your service. A very minor or routine change should not require you to carry out a new risk assessment.
5.5 In Table 11, we have provided guidance on what could amount to a significant change by listing a few indicative scenarios that are likely to amount to a significant change.

Table 11: Guidance on significant change

Type of change
Guidance
Outcome
One overarching question you should consider when evaluating whether a change is significant is
the size of your service. For instance, a relatively minor change on a large service is likely to have a
significant impact, while it could take a much larger change on a smaller service to trigger the
need to review their risk assessment.
Likely to amount to a significant change
Your proposed change is very likely to amount to a significant change on your service if any of the following apply: • The proposed change alters the risk factors which you identified in your last children’s risk assessment. • The proposed change materially impacts a substantial proportion of your child user base or
If yes, you must carry out a new risk assessment relating to this change
changes the kind of users you expect to see on your service, particularly the age groups or volume of children likely to use your service. • The proposed change materially impacts a vulnerable user group, such as children in a certain age group or with certain characteristics. • The proposed change materially impacts the efficacy of the measures you have put in place following your last assessment to reduce the risk of harm to children. • The proposed change materially impacts your revenue model, growth strategy and/or ownership in a way that affects its service design. When considering these statements, you should consider if any of the following apply to the proposed change: • Would the change materially impact children, child user experience or child user behaviour in a way that may affect risks? • Does the change affect the ability or incentives of users to create or share content harmful to children on your service? • Will the change affect child user reporting abilities – particularly if the change impacts the user interface or alters reporting processes? • Does the change include new functionalities or enable children to interact differently? For example, this could include functionalities or features which impact how much children use the service or those which enable adults to search for or contact children. • Does it include changes that would affect content or network recommendations on your service? • Does the change involve a new or different content moderation process or approach? • Does it include changes to your business model in terms of how you generate revenue or your growth strategy?
5.6 Examples of the types of design and operational changes which are likely to amount to a significant change include – but are not limited to – the following: • Significant updates to the design of user-facing algorithms, systems and processes, for example, changing the operation of the recommendation system(s). Key examples include: > Introduction of a new recommender system: This may include a recommender to suggest friends and groups to follow, alongside the existing content feed
recommender. It may also include a complete replacement of the existing content recommender.
> Introduction of a new machine-learning model within the existing recommender system: A service could implement a new machine-learning model to enhance the predictions that are made by a recommender system (e.g., updating any kind of modelling with regards to user age, including inferred or estimated ages of child users). These new or enhanced predictions would, in turn, alter the types of content that children are recommended in different age groups.112 > Changing the ‘goal criteria’ of recommender systems: Changing the overall aims that the service has in mind for those systems, for instance, to maximise the average viewing time, or to promote a certain kind of content to be presented to children (e.g., educational content). > New insights from on-platform testing: Indications that a change to a recommender system, for example, the addition of a child-specific feature or focus, may have a significant impact on the risk of harm arising from content harmful to children.
• Adding or removing functionalities: The children’s risk assessment must assess the impact of functionalities on the risk of harm to children, so adding or removing functionalities – such as sharing content, direct messaging, end-to-end encryption or livestreaming, and especially those that might appeal strongly to children, such as generative artificial intelligence, or autoplay features – must be accounted for in the children’s risk assessment. • Changes to platform content rules or content prioritisation: These may alter the types of content that children encounter and subsequently alter the site’s child user base. This could include, for instance, the decision to alter a threshold around what violent content is prohibited on a service. • Updates to the design of user facing functionalities and features: This could include changing the location or prominence of the reporting function or changing the layout, formatting or wording113 of a safety notification which could impact how children engage with safety measures. • Introducing the use of prompts: Such prompts could alert or remind children about options to change content control settings; the timing and language used in these prompts could impact the choices children make and go on to impact the risk of encountering content harmful to children. • Any acquisition that may change the core product offered to child users: This could include integrating functionality from another service following a product acquisition, and child user migration. • Changes in ownership or investment: These may influence how the service operates (a new owner may have different views on how the service should operate). • Changes in the revenue models: Examples that may affect level of risk include new streams of revenue, a significant change in the factors or key performance indicators
112 This is relevant even where a service provider may not be using age assurance approaches to understand the age of children. 113 In particular, if the comprehensibility of written information is changed, based on the likely reading age of the youngest person permitted to use the service without the consent of a parent or guardian. See our Protection of Children Codes for further detail on the accessibility of such features.
that the service maximises to achieve its revenue goals, or changes in sources of revenue that have a significant impact on the design choices of the service.
• Changes in the service’s growth strategy: For example, changes in growth strategy could affect service design choices or the speed of growth of your child user base, or the number of children in different age groups. • Change in capacity (in terms of number of employees): This could include capacity changes that may affect the number and quality of technical resources to assess and mitigate risk to children on your service.

A1 Appendix A: Examples of how to use the Risk Level Table

A1.1 This sub-section includes hypothetical examples to help services providers to understand how to use the Risk Level Table to make accurate judgements about risk levels, as part of their risk assessment of content harmful to children.
A1.2 These examples are not intended to be conclusive risk assessment evaluations for the specific content identified – they are not based on real, evidence-based judgements. For example, they do not explain the accuracy of information; the volume of reporting; the rating for likelihood and impact of the harm; nor do they explain the degree of confidence in the effectiveness of any measures. Service providers should not seek parity or comparison with these illustrations to justify their own risk assessments.
A1.3 In addition, these examples do not cover service providers’ obligations under data protection legislation. When considering user data, we encourage providers to familiarise themselves with the data protection legislation, and to consult the ICO’s guidance on UK GDPR requirements and the Children’s Code.114
Boxes 4-9: Risk Level Table theoretical examples
Content which encourages, promotes, or provides instructions for an eating disorder or
behaviours associated with an eating disorder – high risk
A discussion forum is intended to provide support for individuals suffering from eating disorders.
It has 50,000 monthly UK users and uses self-declaration (without more) on entry to the site to
apply its minimum age requirement of 16 years old. Users can post content including text, images
and videos. Users can also search for and comment on other users’ content.
The service has a user reporting system, and all reported content is reviewed by human
moderators. The forum’s terms of service set out that content promoting or providing instructions
for an eating disorder is prohibited. However, the service has limited resources and it can take
some time before content that goes against the forum’s terms of service is removed.
User reports indicate that a substantial amount of content encouraging or promoting eating
disorders is present. The provider cannot rule out that more instances of this content may be
occurring that are not reported by users, especially as it has some of the risk factors associated
with content promoting eating disorders, such as searching, posting and commenting on content.
The service recognises self-declaration can be bypassed easily, and so assumes that children
under their minimum age are likely to access the site and may encounter content promoting or
encouraging eating disorders.
Despite being a relatively small service, the provider concludes it is high risk for content that
encourages or promotes eating disorders, especially considering that the content could have a
severe impact on child users – both for 16-17-year-olds permitted on the service and younger
children not permitted on their service.
Content which encourages, promotes, or provides instructions for an eating disorder or
behaviours associated with an eating disorder – negligible or no risk
A large vertical search service specialises in travel searches, including for flights and hotels. It has
around 10 million monthly UK users. The service does not use a highly effective age assurance
process, but has commissioned independent market research and consulted other third-party
sources which indicate the large majority of users are likely to be adults.
It has a user reporting system which could be used to report content such as eating disorder
content, should such content feature in search results.
Though the service’s publicly available statement does not prohibit this kind of content, there has
never been any evidence or suggestion of content which encourages, promotes or provides
instructions for an eating disorder appearing in search results, and the provider can see no way in
which this could ever happen, due to the nature of how its search results are sourced. Even
though it is a large service, and children may access the service, the provider concludes it has
negligible or no risk for this kind of content.
Content which encourages, promotes or provides instructions for an act of serious violence
against a person – high risk
A large video-sharing platform has around 10 million monthly UK users. The platform uses self-
declaration and inferences from user data to apply its minimum age requirement of 13 years old.
Using these methods, it estimates that the service is likely to have more than 1 million monthly UK
users who are children between the ages of 13 and 18. This evidence is not robust enough to rule
out that children under the age of 13 are present for the purposes of the children’s risk
assessment, and they conclude that children under 13 use the service. Users can livestream and
can post videos with anonymous or pseudonymous user profiles. The platform has recommender
systems and autoplay functionality that suggest content and new user connections.
The service prohibits violent content in its terms of service, and has extensive and sophisticated
safety measures in place to enforce this. These include automated content moderation systems
and blocking or restricting users found to post violative content. It has a child-friendly reporting
system and works with various trusted flaggers.
The service has a sophisticated process for estimating the ‘harm prevalence’ rate which is the
percentage of views on the service that are of content violating its terms of service. While its
harm prevalence rate for violent content is low in percentage terms, given the high number of
child users and views, the number of children being impacted by violent content is still significant.
This is consistent with their user reports, including child user reports, indicating many incidents of
violent content on the service.
Despite relevant existing controls, the provider concludes the service is high risk for content which
encourages, promotes or provides instructions for an act of serious violence. This is based on the
evidence of the volume of content promoting serious violence and the number of children across
different age groups using the service.
Content which encourages, promotes or provides instructions for an act of serious violence
against a person – medium risk
An online game involves teams competing. Users can communicate with their teammates through
group messaging and voice chat functionality provided by the service while playing. Teams can be
formed from within user groups or with non-connected users.
The service has over 500,000 monthly UK users. To apply its minimum age requirement of 13
years old in practice the service uses a highly effective age assurance process which age checks all
users. The service estimates that there is a moderate number of children aged between 13 and 18
on the service, so there is some scope for this kind of content to impact many children.
The service prohibits content that encourages or promotes acts of serious violence against people
in its terms of service, and allows a user to report another user for this kind of behaviour. Users
are warned or banned for using language which promotes or encourages acts of serious violence
in these cases. The service encourages users to use the reporting system, including reassuring
them that if they report someone, that person will not be informed by the provider who reported
them. It also employs product nudges such as a pop-up prompt before a message is sent, when it
detects a potentially violent message being written.
From user reports, including child user reports, the service has some evidence of content
encouraging serious violence against a person, for example, another user within chats, and bans
some users for this reason. The provider is aware that there may also be unreported violent
content. While the number of children directly affected by each incident is small, there are
multiple incidents each month and the provider assesses that the impact on those affected is
likely to be moderate. The provider concludes the service is medium risk for content encouraging,
promoting, or providing instructions for an act of serious violence against a person.
Content that promotes depression, hopelessness and despair (‘depression content’) – high risk
A large social media service has over 10 million monthly UK users. Users can post content, including text, images and videos, and users are recommended content in a personalised ‘feed’, which has autoplay functionality. Users can also comment on other users’ posts. The site states a minimum age requirement of 13 years old in its terms of service, and relies exclusively on self-declaration on entry to the site. The service is known to be popular amongst users of different ages, and third-party data suggests that there are many child UK users of all ages, meaning there is broad scope for this kind of content to impact many children. As well as a content reporting system, the service allows users to report and block other users. The service does not prohibit depression content in its terms of service. It identifies the presence of this kind of content on the service, as it receives a high volume of user reports of depression content. External research from child behavioural experts they regularly consult has highlighted that this kind of content is extremely common on the service, with severe impacts for children that have encountered in high volumes, primarily through a recommender system. Additionally, as the service does not use highly effective age assurance approaches, the service does not have robust controls or evidence to demonstrate that children under the age of 13 are not on the service, so the provider assesses children below their minimum age are also likely to access the site and encounter this kind of content. The provider concludes the service is high risk for content that promotes depression, hopelessness and despair. As the provider has identified the presence of this kind of non-designated content on the service, they notify Ofcom, noting the kind (‘content that promotes depression, hopeless and despair’), and the incidence of this kind of non-designated content on the service.
Content that promotes depression, hopelessness and despair (‘depression content’) – negligible
A sustainability charity has a website which shares support resources and information about
making sustainable choices when buying food. The service has a discussion forum where users can
share experiences and offer tips to one another. It has 10,000 monthly UK users and has no
measures in place to understand the age of users on the service. Users can share images and text,
and respond to each other’s comments.
The terms of service prohibit users from posting content that promotes depression, alongside
other potentially harmful content. Staff at the charity review all content and comment boxes after
users have posted the content. It also has a complaint system and looks at any complaints
promptly.
The service has a few of the risk factors for depression content, but as they review all content and
comments shared by users, it would be likely to know if there had ever been such content on the
service, and there has never been any. Additionally, there have never been any complaints about
depression content from users. The provider concludes it presents negligible risk for depression
content.
(1) This section sets out the duties about risk assessments which apply in relation to all regulated user-to-user services.
(1) This section sets out the duties about risk
assessments which apply in relation to
regulated user-to-user services that are likely to
be accessed by children (in addition to the
duties about risk assessments set out in section
9 and, in the case of services likely to be
accessed by children which are Category 1
services, the duties about assessments set out
in section 14).
(3) A duty to take appropriate steps to keep an
(3) A duty to take appropriate steps to keep a
illegal content risk assessment up to date,
children’s risk assessment up to date, including
including when Ofcom make any significant
when Ofcom make any significant change to a
change to a risk profile that relates to services
risk profile that relates to services of the kind in
of the kind in question.
question
(5) Where a children’s risk assessment of a
service identifies the presence of non-
designated content that is harmful to children,
a duty to notify Ofcom of – (a) the kinds of such
content identified, and (b) the incidence of
those kinds of content on the service.
(c) the level of risk of the service being used for the commission or facilitation of a priority offence;
(c) the level of risk of harm to children
presented by different kinds of content that is
harmful to children, giving separate
consideration to children in different age
groups;
(e) the level of risk of functionalities of the service facilitating the presence or dissemination of illegal content or the use of the service for the commission or facilitation of a priority offence, identifying and assessing those functionalities that present higher levels of risk;
(e) the extent to which the design of the
service, in particular its functionalities, affects
the level of risk of harm that might be suffered
by children, identifying and assessing those
functionalities that present higher levels of risk,
including functionalities – (i) enabling adults to
search for other users of the service (including
children), or (ii) enabling adults to contact other
users (including children) by means of the
service;
(g) the nature, and severity, of the harm that might be suffered by individuals from the matters identified in accordance with paragraphs (b) to (f);
(g) the nature, and severity, of the harm that
might be suffered by children from the matters
identified in accordance with paragraphs (b) to
(f), giving separate consideration to children in
different age groups;
(6) In this section references to risk profiles are
(7) In this section references to risk profiles are
to the risk profiles for the time being published
to the risk profiles for the time being published
under section 98 which relate to the risk of
under section 98 which relate to the risk of
harm to individuals presented by illegal
harm to children presented by content that is
content.
harmful to children.
(1) This section sets out the duties about risk assessments which apply in relation to all regulated search services.
(1) This section sets out the duties about risk
assessments which apply in relation to
regulated search services that are likely to be
accessed by children (in addition to the duties
about risk assessments set out in section 26).
(3) A duty to take appropriate steps to keep an
(3) A duty to take appropriate steps to keep a
illegal content risk assessment up to date,
children’s risk assessment up to date, including
including when Ofcom make any significant
when Ofcom make any significant change to a
change to a risk profile that relates to services
risk profile that relates to services of the kind in
of the kind in question.
question.
(5) An ‘illegal content risk assessment’ of a
(5) A ‘children’s risk assessment’ of a service of a particular kind means an assessment of the following matters, taking into account the risk profile that relates to services of that kind –
service of a particular kind means an
assessment of the following matters, taking
into account the risk profile that relates to
services of that kind –
(b) the level of risk of functionalities of the
(b) the level of risk of children who are users of
service facilitating individuals encountering
the service encountering search content that is
search content that is illegal content,
harmful to children which particularly affects
identifying and assessing those functionalities
individuals with a certain characteristic or
that present higher levels of risk;
members of a certain group;
(d) how the design and operation of the service
(d) the different ways in which the service is
(including the business model, governance, use
used, including functionalities or other features
of proactive technology, measures to promote
of the service that affect how much children
users’ media literacy and safe use of the
use the service, and the impact of such use on
service, and other systems and processes) may
the level of risk of harm that might be suffered
reduce or increase the risks identified.
by children;
(f) how the design and operation of the service
(including the business model, governance, use
of proactive technology, measures to promote
users’ media literacy and safe use of the
service, and other systems and processes) may
reduce or increase the risks identified.
(7) See also – (a) section 34(2) and (9) (records
(7) See also – (a) section 34(2) and (9) (records
of risk assessments), and (b) schedule 3 (timing
of risk assessments), and (b) schedule 3 (timing
of providers’ assessments).
of providers’ assessments)’
114 ICO, UK GDPR guidance and resources; ICO, Age appropriate design: a code of practice for online services; ICO, ‘Likely to be accessed’ by children guidance.

A2 Appendix B: Comparison of children’s and illegal content risk assessment duties

A1.1 For user-to-user services, the illegal content risk assessment duties can be found in section 9 of the Act, and the children’s risk assessment duties can be found in section 11 of the Act.
Table 12: Comparison of the illegal content risk assessment duties and the children’s risk assessment duties – user-to-user services
Illegal content risk assessment duties Children’s risk assessment duties Section 9 (User-to-user services) Section 11 (User-to-user services)
(2) A duty to carry out a suitable and sufficient (2) A duty to carry out a suitable and sufficient illegal content risk assessment at a time set out children’s risk assessment at a time set out in, in, or as provided by, schedule 3. or as provided by, schedule 3.
(4) Before making any significant change to any (4) Before making any significant change to any aspect of a service’s design or operation, a duty aspect of a service’s design or operation, a duty to carry out a further suitable and sufficient to carry out a further suitable and sufficient illegal content risk assessment relating to the children’s risk assessment relating to the impacts of that proposed change. impacts of that proposed change.
Illegal content risk assessment duties Children’s risk assessment duties Section 9 (User-to-user services) Section 11 (User-to-user services)
(5) An ‘illegal content risk assessment’ of a service of a particular kind means an (6) A ‘children’s risk assessment’ of a service of
assessment of the following matters, taking a particular kind means an assessment of the
into account the risk profile that relates to following matters, taking into account the risk profile that relates to services of that kind – services of that kind –
(a) the user base; (a) the user base, including the number of users who are children in different age groups; (b) the level of risk of children who are users of the service encountering the following by means of the service – (i) each kind of primary (b) the level of risk of individuals who are users priority content that is harmful to children of the service encountering the following by (with each kind separately assessed), (ii) each means of the service – (i) each kind of priority kind of priority content that is harmful to illegal content (with each kind separately children (with each kind separately assessed), assessed), and (ii) other illegal content, taking and (iii) non-designated content that is harmful into account (in particular) algorithms used by to children, giving separate consideration to the service, and how easily, quickly and widely content may be disseminated by means of the children in different age groups, and taking into
service; account (in particular) algorithms used by the service and how easily, quickly and widely content may be disseminated by means of the service; (d) the level of risk of harm to children (d) the level of risk of harm to individuals presented by illegal content of different kinds presented by content that is harmful to
or by the use of the service for the commission children which particularly affects individuals
or facilitation of a priority offence; with a certain characteristic or members of a certain group;
Illegal content risk assessment duties Children’s risk assessment duties Section 9 (User-to-user services) Section 11 (User-to-user services) (f) the different ways in which the service is used, including functionalities or other features (f) the different ways in which the service is of the service that affect how much children used, and the impact of such use on the level of use the service (e.g., a feature that enables risk of harm that might be suffered by individuals; content to play automatically), and the impact of such use on the level of risk of harm that might be suffered by children;
(h) how the design and operation of the service (h) how the design and operation of the service (including the business model, governance, use (including the business model, governance, use of proactive technology, measures to promote of proactive technology, measures to promote users’ media literacy and safe use of the users’ media literacy and safe use of the service, and other systems and processes) may service, and other systems and processes) may reduce or increase the risks identified. reduce or increase the risks identified.
(7) See also – (a) section 23(2) and (10) (records (8) See also – (a) section 23(2) and (10) (records of risk assessments), and (b) schedule 3 (timing of risk assessments), and (b) schedule 3 (timing of providers’ assessments). of providers’ assessments).
5.7 For search services, the illegal content risk assessment duties can be found in section 26 of the Act, and the children’s risk assessment duties can be found in section 28 of the Act.
Table 13: Comparison of the illegal content risk assessment duties and the children’s risk assessment duties – search services
Illegal content risk assessment duties Children’s risk assessment duties Section 26 (Search services) Section 28 (Search services)
(2) A duty to carry out a suitable and sufficient (2) A duty to carry out a suitable and sufficient illegal content risk assessment at a time set out children’s risk assessment at a time set out in, in, or as provided by, schedule 3. or as provided by, schedule 3.
(4) Before making any significant change to any (4) Before making any significant change to any aspect of a service’s design or operation, a duty aspect of a service’s design or operation, a duty to carry out a further suitable and sufficient to carry out a further suitable and sufficient illegal content risk assessment relating to the children’s risk assessment relating to the impacts of that proposed change. impacts of that proposed change.
Illegal content risk assessment duties Children’s risk assessment duties Section 26 (Search services) Section 28 (Search services) (a) the level of risk of children who are users of the service encountering search content of the following kinds – (i) each kind of primary (a) the level of risk of individuals who are users priority content that is harmful to children of the service encountering search content of the following kinds – (i) each kind of priority (with each kind separately assessed), (ii) each
illegal content (with each kind separately kind of priority content that is harmful to
assessed), and (ii) other illegal content, taking children (with each kind separately assessed),
into account (in particular) risks presented by and (iii) non-designated content that is harmful
algorithms used by the service, and the way to children, giving separate consideration to children in different age groups, and taking into that the service indexes, organises and presents account (in particular) risks presented by search results; algorithms used by the service and the way that the service indexes, organises and presents search results; (c) the extent to which the design of the service, in particular its functionalities, affects (c) the nature, and severity, of the harm that the level of risk of harm that might be suffered might be suffered by individuals from the by children, identifying and assessing those matters identified in accordance with functionalities that present higher levels of risk, paragraphs (a) and (b); including a functionality that makes suggestions relating to users’ search requests (predictive search functionality); (e) the nature, and severity, of the harm that might be suffered by children from the matters identified in accordance with paragraphs (a) to (d), giving separate consideration to children in different age groups;
Illegal content risk assessment duties Children’s risk assessment duties Section 26 (Search services) Section 28 (Search services)
(6) In this section references to risk profiles are (6) In this section references to risk profiles are to the risk profiles for the time being published to the risk profiles for the time being published under section 98 which relate to the risk of under section 98 which relate to the risk of harm to individuals presented by illegal harm to children presented by content that is content. harmful to children.