Key facts

  • Applies from 2 August 2026 — the live date for most organisations.
  • Covers chatbots, AI-generated text/image/audio/video, deepfakes, and emotion-recognition or biometric-categorisation systems.
  • Obligations sit with both providers (build the labelling capability) and deployers (use it correctly).
  • Some exemptions apply, e.g. for AI used purely for editorial assistance where a human retains responsibility.
  • Pair with an AI inventory to identify which of your systems are in scope.

What does Article 50 require?

It requires disclosure that a person is interacting with an AI system (unless obvious from context), and labelling of AI-generated or manipulated audio, image, video or text content — particularly deepfakes that could be mistaken for authentic material.

Who does it apply to?

Providers building AI systems that interact with people or generate synthetic content, and deployers using those systems, such as a business running a customer-facing chatbot or publishing AI-generated marketing content.

What counts as a deepfake under the Act?

AI-generated or manipulated image, audio or video content that resembles existing people, objects, places or events and would falsely appear authentic to a reasonable person.

How to prepare

Identify every AI system that talks to customers or generates content, add a clear disclosure (e.g. "You're chatting with an AI assistant"), and put a labelling process in place for AI-generated media before the 2 August 2026 deadline. See the full EU AI Act timeline for how this fits with other deadlines.

Frequently asked questions

What is Article 50 of the EU AI Act?

Transparency obligations: telling people they're interacting with AI, and labelling AI-generated or manipulated content such as deepfakes.

When does the EU AI Act take effect?

It entered into force in August 2024 and obligations phase in — prohibited practices and GPAI rules already apply, transparency duties from 2 August 2026, and high-risk deferred to 2027/2028.

What is the deadline I should care about most?

For most organisations: 2 August 2026, covering transparency and GPAI governance, plus your high-risk classification ahead of the 2027 deadlines.

Does the EU AI Act apply to internal AI tools?

It can — using AI under your own authority makes you a deployer, and high-risk internal use carries obligations even without selling the AI.

Is a popular AI chat assistant covered by the EU AI Act?

Yes, as general-purpose AI — the provider carries GPAI obligations, and your use as a deployer may trigger transparency duties.

Related pages

Sources

Last updated 19 June 2026.