When does Article 50 of the EU AI Act apply?

Article 50 transparency obligations apply from 2 August 2026. There is one transition: systems that generate synthetic content and were already on the market have until 2 December 2026 to implement machine-readable marking under Article 50(2).

What does Article 50 require?

Four duties: AI systems interacting with people must disclose they are AI; systems generating synthetic audio, image, video or text must mark outputs machine-readably; deepfakes must be visibly labeled; and people exposed to emotion recognition or biometric categorisation must be informed.

What is the penalty for breaching Article 50?

Article 50 breaches fall in the middle penalty tier of Article 99: up to €15 million or 3% of worldwide annual turnover, whichever is higher (for SMEs, the lower of the two).

Are you ready for Article 50?

The EU AI Act's transparency obligations apply from 2 August 2026 — chatbot disclosure, machine-readable marking of AI-generated content, deepfake labels and emotion-recognition notices. Answer seven yes/no questions to get your tailored duty list. Deterministic, no sign-up.

1.Do you operate an AI system that interacts directly with people in the EU (chatbot, voice assistant, AI agent)?
Article 50(1) — users must be informed they are interacting with AI, unless it is obvious from context.
2.Does your system generate synthetic audio, image, video or text content?
Article 50(2) — providers of generative systems must mark outputs as artificially generated.
3.Are those outputs marked in a machine-readable format as AI-generated?
Only relevant if the question above is answered yes.
4.Do you publish deepfakes — AI-generated or manipulated content resembling real people, places or events?
Article 50(4) — deepfakes carry a visible disclosure duty on top of machine-readable marking.
5.Are those deepfakes visibly labeled as artificially generated or manipulated?
Only relevant if the question above is answered yes.
6.Do you use emotion recognition or biometric categorisation systems?
Article 50(3) — subjects must be informed. Workplace and education uses of emotion recognition are prohibited outright (Article 5).
7.Do you inform the people exposed to those systems about their operation?
Only relevant if the question above is answered yes.

Answer the questions above to see your Article 50 duty list. Nothing is stored or sent anywhere.

What Article 50 covers

Article 50 is the EU AI Act's transparency layer — the tier that catches most everyday AI products that are neither prohibited nor high-risk. From 2 August 2026 it requires:

• Art. 50(1) — AI systems interacting with people must disclose they are AI.
• Art. 50(2) — providers of systems generating synthetic audio, image, video or text must mark outputs in a machine-readable format (transition to 2 December 2026 for systems already on the market).
• Art. 50(4) — deepfakes must be visibly labeled as artificially generated.
• Art. 50(3) — people exposed to emotion recognition or biometric categorisation must be informed.

Breaches sit in the €15M / 3% penalty tier — and the same date activates the Commission's GPAI enforcement powers under Article 101. Read what actually changes on 2 August 2026 and the content-marking guide, or run the full risk checker to classify your system across all tiers.