On August 2, the European Commission and national regulators began enforcing the transparency provisions of the European Union’s Artificial Intelligence Act, requiring chatbots, AI-generated media and deepfakes to identify themselves as artificial, according to the Commission’s own announcement.
What the rules require
Under Article 50 of the AI Act, providers and companies deploying AI systems in the EU must now meet four disclosure duties: chatbots and other systems that interact directly with people must tell users they are talking to AI, not a human; tools that generate or edit images, audio or video must embed machine-readable marks so the output can be detected as synthetic; deployers of emotion-recognition or biometric-categorization systems must inform anyone subjected to them; and AI-generated text on matters of public interest must be labeled as artificial unless a human editor has reviewed it.
The Commission adopted formal guidance on July 20 to help providers and the regulators enforcing the law apply it consistently across the bloc’s 27 member states.
Penalties and a partial grace period
Non-compliance can bring fines of up to €15 million or 3% of a company’s worldwide annual turnover, whichever is higher. Most obligations apply immediately to systems already on the market, but providers get until December 2 to add the required marking and detection features to generative AI tools already in use.
More than 180 organizations have already signed the EU’s voluntary Code of Practice on AI-generated content, which the Commission says grants signatories a “presumption of conformity” with the new rules.
The disclosure duties survive unchanged even as other parts of the law have been softened: a Digital Omnibus overhaul finalized earlier this year pushed back the AI Act’s tougher high-risk system rules to December 2027, but left Article 50’s transparency timeline in place. For a fuller rundown of how the law is structured, see our explainer on the EU AI Act.