An Online Safety Act illegal content risk assessment is a written evaluation of how likely it is that illegal content appears on your user-to-user or search service, and how much harm it could cause. Ofcom's method has four steps: understand the kinds of illegal content to assess, assess the risk of each, decide and implement measures from its codes, then report, review and update. You need one within three months of launching an in-scope service, and you must keep a written record.

This guide follows Ofcom's four steps with practical notes on evidence, the codes measures for small and large services, and record-keeping. It is general information, not legal advice. If you are not sure the Act applies, start with does the Online Safety Act apply to my website?

Who must do an illegal content risk assessment, and when

Every regulated user-to-user and search service must carry one out, whatever its size. The duties became enforceable on 17 March 2025. After that:

  • a new in-scope service must complete its assessment within three months of launch
  • you must carry out a new assessment before making a significant change to your service's design or operation, such as adding private messaging, live streaming, uploads or a recommender feed
  • Ofcom recommends reviewing it at least annually, and you must review it when Ofcom makes significant changes to its Risk Profiles

This is separate from the children's risk assessment, which applies if your service is likely to be accessed by children, and from Part 5 age assurance duties for pornography you publish yourself. The Online Safety Act overview maps how they fit together.

Step 1: understand the priority illegal content you must assess

Ofcom's risk assessment guidance lists kinds of priority illegal content that every assessment must cover, 18 at the time of writing, as the list has grown with new priority offences. They include:

  • terrorism, hate, and harassment, stalking, threats and abuse
  • child sexual exploitation and abuse, including grooming and image-based child sexual abuse material (CSAM) and CSAM URLs
  • intimate image abuse, extreme pornography and sexual exploitation of adults
  • human trafficking, unlawful immigration, fraud and financial services offences, proceeds of crime
  • drugs, weapons, encouraging or assisting suicide and serious self-harm, foreign interference, animal cruelty and cyberflashing

You must assess every kind, even if you conclude the risk is negligible, plus any non-priority illegal content likely to occur on your service. For user-to-user services you also consider how the service could be used to commit or facilitate priority offences, not just host the content.

Then use Ofcom's Risk Profiles to identify your risk factors. These are features and characteristics linked to particular harms: user profiles, anonymous accounts, direct messaging, livestreaming, file sharing, hyperlinking, group chats, and business models such as payments between users. Write down which apply to you.

Adult platforms: expect intimate image abuse, extreme pornography, sexual exploitation of adults, human trafficking and CSAM to be live questions for any service where users upload or stream sexual content. Treat them as the core of your assessment, not footnotes.

Step 2: assess the likelihood and impact of each kind

For each kind of illegal content, assess separately how likely it is to occur and how severe the impact would be, then use the risk level tables in Ofcom's guidance to assign one of four levels: negligible, low, medium or high.

Ofcom expects you to use evidence, and to consider design features, algorithms, business model, user base, unintended uses and how effective your existing controls are. Core inputs every service has:

  • your Risk Profile results from Step 1
  • user reports and complaints, and what they were about
  • moderation logs: what you removed and why
  • law enforcement requests and your CSAM reporting and law enforcement history
  • product data such as how many users message strangers, how quickly new accounts post, and where content comes from

Larger services are expected to add enhanced inputs, such as results of content moderation testing, commissioned research and external expert views.

Result across all kindsService classificationConsequence
Negligible or low for everythingLow-risk serviceMinimum set of codes measures
Medium or high for one kindSingle-risk serviceAdditional measures for that harm
Medium or high for two or more kindsMulti-risk serviceFurther governance and moderation measures

Be honest. A service that allows anonymous uploads of explicit video and rates itself negligible for intimate image abuse will not be credible, and an implausible assessment is worse than an honest "medium" with good controls.

Step 3: decide measures using Ofcom's illegal harms codes

Next, work out which measures in Ofcom's illegal content codes of practice apply to you, implement them, and record what you did. The codes are grouped into governance and accountability, content moderation, reporting and complaints, recommender systems, settings and user support, terms of service, user access and user controls.

Measures that apply to all services

  • name an individual accountable to the most senior governance body for compliance with the illegal content, reporting and complaints duties
  • have a content moderation function that reviews suspected illegal content and takes it down swiftly
  • provide easy-to-find and easy-to-use reporting and complaints processes, and act on them
  • have terms of service that explain how users are protected from illegal content, written clearly and accessibly

Measures for large or multi-risk services

"Large" means an average of 7 million or more monthly UK users. Large services and multi-risk services pick up more measures, such as an annual review of risk management by the governance body, written statements of staff responsibilities, internal monitoring and assurance, performance targets for moderation, and training for moderators.

Harm-specific measures

Some measures only apply where you are medium or high risk for a particular harm, such as default settings that stop strangers messaging children where grooming is a risk, or automated detection such as hash matching against known CSAM for certain services at risk of image-based CSAM. Check the measure tables in the codes for the size and risk conditions that trigger each one.

Following the codes measures that apply to you means you are treated as complying with the related duties. You can use alternative measures, but you must record how they meet the duties, which is more work and more risk. For adult platforms, build your content moderation policy around the measures you adopt here.

Step 4: report, review and update

Report the outcome to your governance body. For a small service, Ofcom accepts a senior manager. Then monitor whether your measures work and what risk is left, and update the assessment when your service, your evidence or Ofcom's Risk Profiles change. Put a review date in the calendar now.

Record-keeping: what Ofcom expects to see

You must keep a written record of every risk assessment and of the measures you take. Ofcom publishes a record-keeping template and record-keeping and review guidance. A good record includes:

  1. a description of the service, its features and user base
  2. the risk factors identified from the Risk Profiles
  3. for each kind of illegal content: the evidence considered, likelihood, impact and final risk level, with reasons
  4. the codes measures you adopted, or alternatives and why they are equivalent
  5. who approved it and when, and the next review date

Ofcom does not need the record unless it asks for it or you are a categorised service. But it does ask: in December 2025 it fined a file-sharing service £20,000 for failing to respond to information requests, including a request for its illegal content risk assessment. Keep the document current and know where it is.

Worked example: a small adult creator marketplace

Take a hypothetical platform with around 40,000 monthly UK users, where verified creators upload clips and fans can message creators.

  • Step 1: risk factors include user uploads, direct messaging, payments between users and adult content. Priority kinds needing close attention: intimate image abuse, extreme pornography, CSAM, sexual exploitation of adults, human trafficking, harassment and fraud.
  • Step 2: creator ID and consent checks lower the likelihood of non-consensual uploads, but complaints show occasional reports of leaked content re-uploaded by other creators. The platform assesses intimate image abuse as medium and fraud (fans scammed by fake creator accounts) as medium. With two medium risks, it is a multi-risk service.
  • Step 3: it adopts all baseline measures plus the multi-risk governance measures, a named accountable director, moderator training and a fast-track route for intimate image reports. It records each measure against the code reference.
  • Step 4: the director signs off, monthly moderation figures go to the board, and a review is booked before the planned launch of livestreaming, which counts as a significant change.

Illegal content duties checklist: common mistakes to avoid

  • Skipping kinds that "obviously" do not apply. Every priority kind needs a line of reasoning, even if the answer is negligible. A blank row looks like you never considered it.
  • Assessing the service you plan rather than the one you run. The assessment must reflect your live features, including the old private messaging tool nobody remembers switching on.
  • No evidence trail. "We think the risk is low" is not evidence. Cite complaint volumes, moderation outcomes and checks you run.
  • Measures that exist only on paper. If your record says reports are reviewed within 24 hours, your logs should show it.
  • Forgetting third-party features. Embedded chat widgets, comment plugins and community tools supplied by vendors are still part of your service.

Frequently asked questions

Do I have to send my risk assessment to Ofcom?

Not routinely. You must keep it and produce it if Ofcom asks. Category 1 and 2A services have been asked to supply their records by October 2026 and publish summaries by November 2026.

Can I use a consultant or software tool to do it?

Yes, but the assessment must reflect your own service and evidence, and responsibility stays with you. Ofcom's own digital toolkit is a useful free starting point.

How long does a risk assessment take?

For a small service with simple features, a well-prepared team can produce a credible first version in a few days. Platforms with uploads, messaging and payments should expect longer, mainly to gather evidence.

Next steps

Download Ofcom's template, block out a day with whoever runs product and moderation, and complete Steps 1 and 2 with the evidence you already have. If you need outside help with moderation tooling or age checks, compare providers in our compliance and verification services directory.