AI Labelling Rules 2026: What You Need to Do Now

From 2 August 2026, the AI labelling obligation under Article 50 of the EU AI Act applies. If you run a chatbot on your website, publish AI-generated content or use AI in your content production, transparency becomes mandatory. Here's what that means in practice and how to implement it - hands-on and without legal jargon.

What the AI labelling obligation covers

Article 50 is the transparency part of the EU AI Act. The core idea: people should be able to tell when they are interacting with an AI or looking at AI-generated content. Some duties fall on the providers of the AI systems, others on the deployers - the companies that use an AI on their website. As a website operator, two points matter most: your chatbot and your content.

Chatbots must identify themselves

If you use an AI chatbot, visitors must be able to tell clearly and in good time that they are writing with an artificial intelligence and not a human. This information belongs at the start of the interaction - visible in the chat window, not hidden in the legal notice. A first message like "Hi, I'm an AI assistant" or a persistent "AI" label in the chat meets the requirement. The duty only falls away if it is obvious from the context anyway - but you shouldn't rely on that.

Labelling AI-generated content

If images, audio or video are generated or altered by AI and could be mistaken for real (think deepfakes), they must be marked as artificially created. This does not apply to purely decorative, obviously stylised illustrations - they are clearly recognisable as design and do not depict real people or events. AI-generated text that informs the public on matters of public interest is also subject to a labelling duty; ordinary marketing or product copy usually is not.

The key exemption: editorial review

Content that has been editorially reviewed and for which a person or organisation takes responsibility is largely exempt from the text-labelling duty. That's exactly how we work: AI is a tool for research and drafts, but every published article is reviewed and owned by a human. If you document this cleanly, you're on the safe side.

Provider or deployer - who is responsible?

The EU AI Act distinguishes between providers, who develop and place an AI system on the market, and deployers, who use it. As a company that uses a ready-made chatbot or AI tool on its own website, you are usually the deployer - and your main job is to ensure transparency towards your users. Important: as soon as you offer an AI system under your own name or substantially modify it, you can become a provider yourself and take on wider duties. When in doubt, it is worth checking which role you are in for which tool - and whether your provider already supplies the machine-readable labelling.

Why transparency is also a marketing advantage

The labelling obligation sounds like red tape at first, but it is an opportunity: showing openly where AI is involved and where humans safeguard quality builds trust. Especially for mid-sized companies, honesty is a differentiator. A clearly labelled AI assistant and a short AI-transparency page look more professional than a chatbot pretending to be human.

What you should do now

Check three things: first, whether your chatbot clearly identifies itself as AI. Second, whether AI-generated media on your site is labelled as such. Third, whether there is named editorial responsibility for your content. A short AI-transparency page that discloses where and how you use AI builds extra trust. Breaches can get expensive - the range goes up to EUR 15 million or 3% of annual turnover - which is why it pays to act now. More on this in our service Data Privacy & IT Security. For orientation, see the official text of Article 50 EU AI Act.

This article is not legal advice. For a binding assessment, consult a lawyer or your data protection officer.