Article 50

Might Article 50 transparency duties apply?

Interactive systems, synthetic content, emotion recognition, biometric categorisation, deepfakes, and public-interest text.

What this checks

Interactive systems, synthetic content, emotion recognition, biometric categorisation, deepfakes, and public-interest text.

Who should use this screen

Teams that design or publish an AI system or its output.

Current dates

Article 50 applies from 2026-08-02. Marking transition for earlier systems ends 2026-12-02. Annex III high-risk Chapter III sections 1-3 starts 2027-12-02. Annex I high-risk starts 2028-08-02. Transparency and high-risk stay on separate tracks.

Official sources

Regulation (EU) 2024/1689, Regulation (EU) 2026/1744, Commission Guidelines of 20 July 2026, the Commission FAQ, and the transparency code of practice on EUR-Lex and the Commission site.

Limitations

Decision support only. No high-risk conformity assessment, CE marking, quality-system build, or fundamental-rights workflow is included.

How to read the result

A yes on one listed path means a transparency duty is likely.

A path with no yes and no unknown is recorded as no Article 50 duty identified.

Unknown answers require review.

Separate tracks

Transparency dates and high-risk dates are never mixed in the result.

Limits

Does a clear no set of answers close the legal file?

No. It only means these questions did not identify an Article 50 path.

Does a chatbot count as interaction?

A yes on direct interaction marks a transparency duty as likely.

Are emotion and biometric paths separate?

Yes. Each path is its own question.

Which Commission texts sit behind this screen?

The Article 50 guidelines of 20 July 2026, the FAQ, and the transparency code of practice.