Regulation · European Union
EU AI Act — Article 50
Article 50 requires AI providers and professional deployers to give specified AI interaction notices and mark or disclose synthetic content. Its transparency duties apply, except that an amendment gives providers of generative AI systems marketed before 2 August 2026 until 2 December 2026 to add machine-readable output marking. Breaches can lead to fines.
- In force
- Effective
Updated
What it requires
- Providers must make people aware that they are interacting directly with AI, unless this is obvious from the context.
- Providers must mark synthetic audio, images, video and text in a machine-readable way, with exceptions for standard editing and systems that do not substantially alter input.
- Deployers must disclose image, audio and video deepfakes at first exposure; hidden watermarks alone are insufficient, and creative works have flexible disclosure arrangements.
- Deployers must disclose AI-generated or AI-manipulated public-interest text unless it undergoes human review or editorial control with editorial responsibility.
- Deployers of emotion recognition and biometric categorisation systems must inform exposed people, subject to narrow legally authorised crime-related exceptions.
- Disclosures must be clear, distinguishable and accessible, and provided no later than the first interaction or exposure.
Who it affects
AI providers, including those outside the EU placing systems on the market under their own name or trademark, and professional deployers; purely personal, non-professional use is excluded.
Timeline
Article 50 transparency duties began applying, subject to the narrow extension for machine-readable marking of outputs from previously marketed generative AI systems. Our story