Transparency obligations
Article 50 requires that people are told when they interact with an AI system (e.g. a chatbot), when content is AI-generated or manipulated (deepfakes), and when emotion-recognition or biometric-categorisation systems are used.
Article 50 contains four distinct duties. Providers must design systems that interact directly with people so those people are informed they are dealing with AI, unless it is obvious. Providers of generative systems must mark synthetic audio, image, video or text in a machine-readable, detectable format. Deployers of emotion-recognition or biometric-categorisation systems must inform the people exposed to them. And deployers must disclose deepfakes, and disclose AI-generated or manipulated text published to inform the public on matters of public interest unless it has undergone human editorial review. Disclosure must be given at the first interaction, clearly and accessibly.
- Art. 50
Transparency obligations: why it matters
These duties are already enforceable, which puts them ahead of the high-risk regime in any realistic compliance sequence. They are also the only AI Act obligations most consumer-facing companies will ever have — and the machine-readable marking duty is an engineering commitment rather than a copy change, requiring provenance metadata or watermarking in the generation pipeline, robust enough to survive ordinary processing.
Transparency obligations in practice
A news publisher generates article summaries with an LLM. If a human editor reviews and takes editorial responsibility before publication, the Article 50 text-disclosure duty does not bite. If summaries publish automatically, they must be disclosed as AI-generated. The same publisher's AI voice reader must be marked as synthetic audio regardless, because that duty sits on the provider of the generative system and has no editorial-review exception.
Transparency obligations: the mistake people make
Burying disclosure in the terms of service or a privacy policy. Article 50 requires the information to be provided at the latest at the time of the first interaction or exposure, clearly and distinguishably, and in a form accessible to people with disabilities — a footer link is not first-interaction disclosure.
Transparency obligations: common questions
- Do we have to label AI-generated content?
- Yes. Providers of generative AI systems must mark outputs in a machine-readable format detectable as artificially generated or manipulated. Deployers must additionally disclose deepfakes and, in some cases, AI-generated text published to inform the public.
- Is there an exception for human-reviewed content?
- For AI-generated text published to inform the public on matters of public interest, the disclosure duty does not apply where the content has undergone human review or editorial control and a natural or legal person holds editorial responsibility.
- Do transparency duties replace high-risk obligations?
- No. They apply in addition. A high-risk system that also interacts with people or generates synthetic content must satisfy Chapter III and Article 50 together.
Transparency obligations: terms that come up alongside it
Run the free 2-minute EU AI Act checker — no sign-up.