From 2 August 2026, anyone deploying an AI system that generates or manipulates image, audio or video content amounting to a deepfake must disclose that the content was artificially generated — even without intent to deceive, and even where no real person is depicted. People interacting with a chatbot must also be told they're talking to a machine.
- Article 50 of the EU AI Act became applicable on 2 August 2026.
- The disclosure duty applies without any intent to deceive, and even where no real person is depicted.
- AI-generated output must be machine-readable as such; chatbot users must be told they're talking to a machine.
- This is EU law, not UK law — but it applies to any UK business putting AI-generated marketing into the EU market.
What Article 50 actually requires
The European Commission's guidance on Article 50 sets out two distinct obligations that matter to small businesses. First, deepfake content — image, audio or video generated or manipulated by AI to resemble real people, objects, places or events — must be labelled as artificially generated or manipulated. Second, AI-generated text published on matters of public interest, and AI chatbots, must make clear that a person is dealing with a machine.
Why this reaches UK businesses at all
Article 50 is EU law. It does not directly bind a UK-only business with no EU customers. But it applies to anyone putting AI-generated content into the EU market — which in practice covers UK businesses running video ads, product demos or outreach that reaches EU audiences on social platforms. The Commission has also published a Code of Practice on Transparency of AI-Generated Content, which it has assessed as adequate for demonstrating compliance with the marking and labelling obligations.
The practical question this leaves you with
The story leaves a small business reader with a concrete question: do I have to say my video is AI? If you're using an AI avatar presenter, the honest answer is generally yes for EU-facing content, and the sensible default is to label it regardless of audience — it costs nothing and removes the ambiguity. The same logic applies to AI meeting bots that record and summarise client calls: participants should be told a bot is recording and transcribing, before disclosure rules force the point.
Frequently asked questions
Do I have to label AI-generated video content under UK law?
Not under UK law directly — Article 50 is an EU AI Act requirement. It applies to UK businesses whose AI-generated content reaches the EU market.
Does the labelling rule apply if no real person appears in the video?
Yes. The Commission's guidance is explicit that the deepfake disclosure duty applies even where no real person is depicted and without any intent to deceive.
What's the easiest way to comply if I use an AI video presenter or meeting bot?
Use tools that build disclosure into the output by design, and label AI-generated content and AI meeting participants as a default habit rather than a case-by-case judgement call.