Article 50 stopped being a deadline and became present tense
Article 50 of the EU AI Act has applied since 2 August 2026, it was not deferred by the Digital Omnibus, and only the machine-readable marking duty in Article 50(2) still has a grace period, until 2 December 2026.
- In force since
- 2 August 2026
- Reviewed and published
- 24 August 2026
- modules reviewed
- 74
- competencies touched
- 4
Most coverage of 2026 was about postponement. Article 50 is the part that did not move. It is live, it splits duties between providers and deployers, and the people it touches are marketers, communication teams, product owners and anyone running a chatbot.
What changed
- Article 50 has applied since 2 August 2026 and the Digital Omnibus left it untouched, while the standalone Annex III high-risk regime moved to 2 December 2027.
- The Commission published its final guidance on 20 July 2026 and confirmed the date.
- One narrow exception remains: for systems already on the market before 2 August 2026, the machine-readable marking duty in Article 50(2) has a grace period until 2 December 2026.
What it means for your people
- The duty is not one duty. Article 50(1) and 50(2) sit with the provider, while 50(3) and 50(4) sit with the deployer, so the answer to a question like should we label this depends on which role you are in.
- Teams that make content, run campaigns or operate a chatbot are in scope today. This is not a compliance project waiting for a kickoff.
- Because the duty is live, the honest question is no longer whether you are preparing, but whether the people concerned can recognise which of the four situations they are in.
What we changed in the platform
- Anchored Article 50 to four competencies: transparency, generative AI, AI Act basics and the AI Act timeline.
- Reviewed the 74 modules those four competencies serve, with the role boundary between provider and deployer made explicit in each of them.
- Gave six sectors their own transparency carrier rather than one generic module, so the example a learner sees comes from their own working day.
Obligations this entry is anchored to
Article 50: transparency
praxikon:eu:ai-act:obligation:article-50-transparency
sha256 239fbac0e4728dc239352b2f88b199c3cc098082ff72e2972b4b5e8a2b121406
Each obligation carries the payload hash of the object version this entry was written against, so the claim can be verified rather than trusted. Implementation graph release: 2.1.0.
Questions people ask about this change
- Was Article 50 postponed along with the high-risk rules?
- No. The Digital Omnibus moved most standalone Annex III high-risk obligations to 2 December 2027 and left Article 50 in place. Article 50 has applied since 2 August 2026.
- Does all AI-generated content have to carry a visible label?
- No, and that shortcut is where most summaries go wrong. Article 50 sets specific notification, marking and labelling duties for particular situations and particular roles, such as direct interaction with an AI system, deepfakes and certain text published to inform the public.
Official sources
See which of your roles this reaches
The 5-minute scan maps your AI use to the people around it and shows where evidence is still thin.