EU tightens AI transparency rules

The European Commission on May 8 published draft guidelines for the EU AI Act’s transparency rules, opening a consultation before the obligations begin to apply on August 2, 2026. The draft covers Article 50 of Regulation (EU) 2024/1689, the part of the law that requires disclosures for certain AI systems even when they are not classified as high-risk. The Commission said the guidance is meant to help providers, deployers and national authorities apply the rules “in a consistent, effective and uniform manner.” ### Which AI systems are covered by these transparency rules? Article 50 applies to several categories of systems that ordinary users are likely to encounter, not only to the AI tools listed elsewhere in the Act as high-risk. The legal text sets disclosure duties for AI systems intended to interact directly with natural persons, certain emotion-recognition or biometric categorisation systems, and AI systems that generate or manipulate image, audio, video or text content in ways that could resemble authentic material. (digital-strategy.ec.europa.eu) The Commission’s draft guidelines say providers of AI systems that interact with people must ensure users are informed that they are dealing with AI, unless that is obvious from the context and from a reasonably well-informed, observant and circumspect user’s perspective. The same guidance says deployers of emotion-recognition and biometric categorisation systems must inform exposed persons when those systems are in use, subject to the exceptions set out in the Act. (eur-lex.europa.eu) ### What exactly will chatbots and consumer AI tools have to tell users? The Commission said providers will have to inform people when they are interacting with an AI system. In practice, that means chatbot-style tools and other conversational systems will need a disclosure unless the artificial nature of the interaction is already apparent. (digital-strategy.ec.europa.eu) The draft guidelines accompany that legal duty with examples and implementation notes for providers and deployers. The Commission’s FAQ says the transparency work includes both the Article 50 guidelines and a separate Code of Practice on transparent generative AI systems. ### What does the EU mean by “machine-readable” synthetic content? (digital-strategy.ec.europa.eu) Article 50 requires providers of certain AI systems to ensure AI-generated or manipulated content is marked in a machine-readable format and detectable as artificially generated or manipulated. The Commission repeated that point in its May 2026 announcement on the draft guidelines. (digital-strategy.ec.europa.eu) The Commission has also been developing a Code of Practice on marking and labelling AI-generated content. A first draft of that code was published in line with a timetable to finalize it in June 2026, according to the Commission. ### Are these rules already in force? (eur-lex.europa.eu) The AI Act entered into force on August 1, 2024, but Article 50 applies later. The Commission and the AI Act implementation timeline both say the transparency obligations become applicable on August 2, 2026. (digital-strategy.ec.europa.eu) The phased rollout matters because companies are being asked to prepare compliance systems before the formal application date. The Commission said its draft guidance is intended to support that transition and to gather stakeholder feedback before adoption. (digital-strategy.ec.europa.eu) ### What happens next before the August deadline? The Commission’s consultation on the draft Article 50 guidelines runs until June 3, 2026. The agency said the feedback will inform the final version of the guidance. June 2026 is also the target date the Commission gave for finalizing the Code of Practice on marking and labelling AI-generated content. (digital-strategy.ec.europa.eu) Those steps will unfold before Article 50 starts applying on August 2, 2026, with providers, deployers and national authorities then expected to use the final guidance in implementation. (digital-strategy.ec.europa.eu)