Ofcom said user-made chatbots are regulated online content

Ofcom's open letter says generative AI chatbots and user-built bots count as user-generated content under the Online Safety Act.

Historical event

November 8, 2024

First source published

November 8, 2024

Site publication

September 18, 2026

What happened

On 8 November 2024 Ofcom, the UK's online-safety regulator, published an open letter to online service providers on how the Online Safety Act 2023 applies to generative AI and chatbots. The letter followed what it called multiple incidents of online harm, including the death of a teenager in the United States who had formed a relationship with a chatbot, and a separate case in which users had built chatbots impersonating real and deceased children without the families' consent. Ofcom wrote the letter to remind providers of duties it said already applied, not to announce a new rule.

What the documents show

The letter states that a site or app is a user-to-user service under the Act if it lets users share chatbot-generated text, images or video with each other, including group-chat functionality and platforms that let users build their own chatbot personas for others to interact with; such AI-generated content is user-generated content, regulated the same way as content a human posted. Separately, a generative AI tool that searches more than one website or database is a search service, and a tool that can generate pornographic material must use highly effective age assurance. The Act's own text confirms that Part 3 ties duties on risk assessment, codes of practice and children's-access assessment to regulated user-to-user and search services, matching the structure Ofcom's letter describes.

The mechanism

The letter does not create a chatbot-specific rulebook; it applies an existing content-and-service classification to a new kind of content. The operative distinction is not whether a message was written by a human or generated by AI, but whether the service lets users share it with others, so a companion app where only the operator's model talks to a single user sits differently under this framework than a platform with a shared library of user-built chatbot personas. Ofcom pairs this with the general duties already due under Part 3: risk assessment of harmful content, a named person accountable for compliance, resourced content moderation, and age assurance where pornographic generation is possible.

What it leaves open

The letter is a compliance reminder, not a finding against any named company, and it does not resolve how a single-user companion chatbot that never shares output with others should be classified if it is not also a search or pornography-generating service. Ofcom says duties phase in as it publishes final codes of practice, so the compliance bar was still being finalised after the letter's date. This is an editorial reading of the letter's scope, not a claim Ofcom makes about companion apps specifically.

• Does the product let users share AI-generated personas or output with other users, or keep every conversation private to one user and the model?
• Has the provider completed the risk assessment and codes-of-practice steps Ofcom's letter says are mandatory?
• Could the product generate pornographic material, triggering the Act's separate age-assurance duty?

Ofcom's letter is useful mainly for what it refuses to do: rather than write new rules for chatbots, it insists existing service categories already reach them, putting the classification question, user-to-user, search, or neither, at the centre of how a companion app is regulated in the UK.

Sources & reading trail

Full text of Ofcom's letter: which chatbot and generative-AI services are user-to-user, search or pornography-generating services, and what duties apply.

Source published: 8 November 2024 · Retrieved: 16 September 2026

Confirms the statutory structure of duties (risk assessment, children's-access assessment, codes of practice) that Ofcom's letter applies to chatbot services.

Source published: Not established · Retrieved: 16 September 2026

Company documents, filings, studies and official records establish the record; the reading and the questions are Lovebot Journal editorial analysis. This retrospective draft does not imply the site published on the event date.

Continue reading

• A wrongful-death suit treated a companion app's persona as a product
• Age assurance guidance separates declaring from proving
• Companion content filters are a policy choice, not a limit
• Browse the complete the archive

Published September 18, 2026, not on the date of the event described.