Skip to content
C50 Clause50EU AI Act transparency — made auditable.EU AI Act evidence, made auditable
Resources · Blog

What the Article 50 deadline actually changed

2 August 2026 came and went without a single headline enforcement action — which is exactly why it is worth writing down what actually changed that day. The obligation did not appear from nothing: it had been written into the regulation since August 2024. What changed is that it became enforceable.

Before and after 2 August 2026

Regulation (EU) 2024/1689 entered into force on 1 August 2024. Article 50's transparency duties were written in from the start, but Article 113 staggered when each chapter starts applying — Article 50 landed in the same wave as the general-purpose-AI-model obligations, 24 months after entry into force, which is 2 August 2026. Before that date the duty existed on paper; after it, a regulator can act on it.

Nothing about the four duties themselves changed on 2 August. What changed is enforceability — and, just as importantly, that the evidence an organisation would need to defend an inspection now has to already exist rather than be produced on request.

Why it was quiet, and why that will not last

National market-surveillance authorities were themselves still standing up in the weeks around the deadline — several member states had not finished designating theirs. A quiet first month is not evidence of low risk; it is evidence of an enforcement apparatus that was still being assembled. The duty did not become optional because nobody enforced it in week one.

What actually changed, operationally

  1. A chatbot or voice assistant that does not disclose it is AI can now be the subject of a complaint, not just a theoretical gap.
  2. Synthetic content — audio, image, video or text — generated without machine-readable marking is a live 50(2) exposure rather than a future one.
  3. A deployer running emotion-recognition or biometric-categorisation systems without telling the people exposed to it is now answerable for that, not just for the underlying data-protection duty.
  4. Deepfake and AI-generated-news disclosures under 50(4) are live for any deployer publishing that kind of content to the EU.

What to actually check now

The honest first question is not “are we compliant” — it is “can we prove what our disclosures said, and when”. A screenshot taken today proves what a page looks like today, not what it said on 2 August. If that has not been recorded continuously, the gap in the evidence starts now, not retroactively.

Key takeaways

  • The four Article 50 duties did not change on 2 August 2026 — their enforceability did.
  • A quiet first month from regulators reflects new enforcement machinery standing up, not a grace period.
  • The evidence question is now live: can you show what your disclosures said, dated, before the deadline — not just what they say today.

The 2-minute applicability check tells you which of the four duties land on your system, and the site scanner checks a live URL against them today.

Evidence, not screenshots

Clause50 turns disclosures like these into versioned, hash-chained, auditor-ready evidence — dated when it happens, never edited afterwards.

Not yet reviewed by counsel. This page explains our reading of Regulation (EU) 2024/1689 and is kept deliberately close to the regulation’s own wording, but it has not been through legal review. Clause50 produces compliance documentation; it is not legal advice and does not by itself make any system compliant — the obligations described here remain yours. Verify anything you rely on against the primary regulation or your own counsel. See our terms.