Ofcom has eight days left to publish the report Parliament gave it three years to write. Its own newest research notice says the study feeding that report will not finish until after the deadline passes.
Estimated reading time: 6 minutes
In short. Section 63 of the Online Safety Act requires Ofcom's first statutory report on content harmful to children "before the end of the period of three years beginning with the day on which this Act is passed" [2]; the Act was given Royal Assent on 26 October 2023, which Ofcom's own page states puts the deadline at "26 October 2026" [1], eight days from this check. Ofcom's own transparency notice on the research feeding that report, last updated 14 October 2026, says its AI chatbot study is "intended to feed into the Content Harmful to Children Review and Report" required "in line with... section 63", but "we expect this research to conclude by the end of this calendar year, and our findings will be published shortly thereafter" [3], weeks after the statutory deadline it is meant to serve. The same page carries an unlabelled update, dated only "September 2026", reporting that a comparable study has "completed" with findings not due "until the first half of 2027" [3]. Two documented mistakes: four stars.
Ofcom opened a call for evidence on 10 February 2026 "to support" its first statutory report under section 63 of the Online Safety Act, stating plainly on its own consultation page: "The Online Safety Act places a duty on Ofcom to publish a statutory report on content that is harmful to children at least every three years, with the first report due by 26 October 2026" [1]. That call closed on 10 March 2026, and the page's status still reads "Pending Statement" as of its last update, 21 April 2026 [1].
The underlying duty is not a target Ofcom set for itself. Section 63(1) requires Ofcom to review the incidence of content harmful to children on regulated services and the severity of harm children suffer as a result, section 63(2) requires a published report on each review, and section 63(5) fixes the first report's deadline at three years from the day the Act was passed [2]. The Act itself records the date it was passed in its opening lines: "26th October 2023" [2]. Three years on from that date is 26 October 2026.
The research the report depends on
Ofcom's evidence base for that report is not limited to the April 2025 Children's Register of Risks the call for evidence says it will build on [1]. A separate page, titled "Transparency notice: Research on AI chatbot, social media and video sharing platforms to explore online safety" and first published 22 June 2026, sets out fresh fieldwork Ofcom says is "intended to feed into the Content Harmful to Children Review and Report that will be prepared and published by Ofcom in line with the requirements of section 63 of the Online Safety Act" [3].
That page, last updated 14 October 2026, four days before this check, describes researchers creating accounts on AI chatbot services including ChatGPT, Grok, SnapMyAI and Talkie, using both 13-17 and 18+ personas, to examine "whether and how harmful content may be generated for different user profiles" [3]. Its own timeline, in its own words: "We expect this research to conclude by the end of this calendar year, and our findings will be published shortly thereafter" [3].
The same page also carries a second, separately described programme, the "Safer Feeds" research, examining what social media and video sharing services recommend to accounts representing 13-17 year olds, which Ofcom says began "summer 2026" [3]. And at the top of the page sits a third block, headed only "Update: September 2026", with no sub-heading or cross-reference saying which of the two named programmes it updates: "We have completed this research, and intend to publish our findings in the first half of 2027 (publication date may be subject to change). As part of carrying out this research, we focused on whether 'sexualised content'... was generated for different user profiles, instead of harmful content" [3]. The word "generated" appears nowhere in the Safer Feeds description, which talks only of content "recommended or otherwise presented"; it appears only in the chatbot notice's own closing line. Whichever study the update describes, it is one of the two programmes this same page says feeds the section 63 report.
The claims, tested
| The claim | What we found | Verdict |
|---|---|---|
| Ofcom's own page: the first section 63 report is "due by 26 October 2026" [1] | Section 63(5) sets the deadline at three years from Royal Assent; the Act records Royal Assent as 26 October 2023 [2] | Accurate. The deadline is real, fixed in law, and eight days away as of this check |
| The AI chatbot research "is intended to feed into" that report [3] | The same notice expects the research to "conclude by the end of this calendar year", two months after the deadline it is meant to serve, with publication only "shortly thereafter" that [3] | A feeder study for a report due in eight days is not expected to finish for at least another ten weeks |
| The page's "Update: September 2026" box reports a completed study with findings due "the first half of 2027" [3] | No heading or cross-reference on the page says which of the two named research programmes this update belongs to | A reader cannot tell which section 63 input has already slipped by up to eight months, only that one has |
The mistakes, counted
A feeder study for a report due in eight days will not even finish for ten more weeks (1). Ofcom's own words tie its AI chatbot research directly to the section 63 report [3], and its own words put the statutory deadline for that report at 26 October 2026 [1]. A study the regulator says will not conclude until "the end of this calendar year", with findings published only "shortly thereafter", cannot have informed a report due weeks earlier.
The page's own update on a completed, delayed study does not say which programme it is updating (2). The "Update: September 2026" box sits as a free-standing item alongside, not nested beneath, either the chatbot notice or the Safer Feeds notice [3]. Its own description, a study on whether content was "generated... for different user profiles", echoes the chatbot notice's wording rather than the Safer Feeds notice's language of content "recommended or otherwise presented". A transparency notice that leaves its own most recent update unattributed is not transparent about the one thing a reader most needs to know: which commitment has already slipped, and by how much.
Credit where due
Ofcom opened its call for evidence for the first section 63 report nine months before the deadline, not at the last minute, and published a discussion resource to help other organisations engage children on a topic it describes, correctly, as potentially upsetting [1]. The deadline has not yet passed as of this check, eight days remain, and nothing here shows the statutory report itself will miss it. The AI chatbot research notice is also candid about its own limits, naming the exact services it tests and the exact age bands of the fictional accounts it uses, rather than describing its methods vaguely [3].
Verdict
Four stars, from two documented mistakes. Parliament gave Ofcom three years to produce its first statutory word on what is harmful to children online, and that clock runs out on 26 October 2026. Ofcom's own newest research notice, updated four days before this check, describes one of the studies feeding that report as unfinished until after the deadline, and reports a second, comparable study as already finished with its findings delayed to some point in the first half of next year, without saying plainly which study that is. A report written before its own evidence has arrived is not the kind of homework this site expects a regulator to hand in.
Sources
- Statutory report on content harmful to children, Ofcom, published 10 February 2026, last updated 21 April 2026, checked 18 October 2026.
- Online Safety Act 2023, section 63, legislation.gov.uk, checked 18 October 2026.
- Transparency notice: Research on AI chatbot, social media and video sharing platforms to explore online safety, Ofcom, published 22 June 2026, last updated 14 October 2026, checked 18 October 2026.