EU AI Act Transparency Rules Are Live: Chatbots Must Now Identify Themselves

Since 2 August 2026, every chatbot operating in Europe has had a new legal obligation: tell the person on the other end that it is not a person. The same goes for deepfakes, AI-generated images, and any content that has been substantially edited by a machine. If you have interacted with an AI system in the EU over the past week, it should have introduced itself.

What the EU AI Act Transparency Rules Require

The transparency provisions, which form part of the broader EU AI Act, impose three core requirements on companies deploying AI in Europe.

First, interactive AI systems like chatbots must disclose their non-human nature before a conversation begins. Users need to know they are talking to software, not a support agent or sales rep.

Second, deepfakes and AI-generated media must carry visible labels. The intent is straightforward: if a video, image, or audio clip was created or significantly altered by AI, the audience should know.

Third, machine-generated or machine-edited content must include machine-readable watermarks. This is not about what users see on screen. It is about giving detection tools, fact-checkers, and platforms the metadata they need to trace content back to its origin automatically.

The AI Office has published a voluntary Code of Practice alongside the mandatory rules, including a set of standardised icons that providers can use to flag AI-generated content. Think of them as the equivalent of a nutrition label, but for whether a human was involved.

What Happens If Companies Ignore This

The enforcement teeth are real. Non-compliance can trigger fines of up to 15 million euros or 3 percent of worldwide annual turnover, whichever is higher. For a company pulling in 10 billion euros a year, that ceiling sits at 300 million euros.

The rules also require explicit warnings when AI is used to analyse someone's emotions or categorise them using biometric data. That covers everything from sentiment analysis in customer service calls to facial recognition systems sorting people by demographic traits.

Why This Matters Beyond Europe

This is the first binding requirement anywhere in the world for AI systems to identify themselves at scale. The EU has a track record of exporting its regulatory frameworks. GDPR reshaped global data privacy practices. The Digital Markets Act forced platform changes worldwide. The AI Act transparency rules are likely to follow the same trajectory.

For companies building or deploying AI tools, the compliance question is no longer theoretical. If your product serves European users, these rules apply now, not in some future implementation window. And if history is any guide, similar requirements will surface in other jurisdictions within the next 12 to 18 months.

The conversation about AI safety and responsibility just got its first concrete regulatory teeth. Whether other governments follow Europe's lead or chart their own course, the era of anonymous AI interaction is ending.