Today, AI transparency becomes a legal duty

Subject: Today, AI transparency becomes a legal duty

Preview: Article 50 of the AI Act applies to interactions, synthetic content and deepfakes.

From 2 August 2026, the transparency obligations in Article 50 of the EU AI Act apply. Organisations in scope must, among other duties, inform people when they are directly interacting with an AI system unless that fact is obvious in context.

Providers of generative systems must add machine-readable marking to certain synthetic outputs. Deployers must disclose deepfakes, certain public-interest text produced without human review or editorial control, and exposure to emotion recognition or biometric categorisation.

Compliance is no longer only a general policy statement. It reaches the interface, content workflow, metadata and evidence that the organisation informed people correctly.

The rules contain exceptions and distinctions between providers and deployers. A limited transition may apply to marking obligations for some existing systems. This briefing is not a substitute for legal advice on your specific use case.

Inventory chatbots, avatars, generation tools, deepfakes, public-interest content and biometric systems. For each, record the owner, required notice, marking method and evidence of control.

AI transparency is no longer only good communication practice; it is becoming a design and operating requirement.