The high-risk deadlines moved in July 2026. Most other obligations did not. Here is what applies when, and who it lands on.
Every provider and deployer. You take measures that support AI literacy among staff working with AI, matched to their role and context. You do not have to guarantee a knowledge level per employee.
Everyone. Eight categories, including social scoring, manipulative AI and emotion recognition at work and in education.
Providers of GPAI models. Supervised by the Commission and the AI Office, not by national authorities (Article 88).
Anyone running a chatbot or publishing AI-generated content. This one was not postponed.
A new prohibition added by the Digital Omnibus. Applies to providers and deployers alike, for images, video and audio.
Member States. Participation is voluntary; SMEs and start-ups get priority and free access.
Moved from 2 August 2026 by the Digital Omnibus.
Standalone systems in hiring, credit scoring, education, critical infrastructure, law enforcement and migration.
Moved from 2 August 2026 by the Digital Omnibus.
AI embedded in products already covered by EU product law: medical devices, machinery, toys.
Moved from 2 August 2027 by the Digital Omnibus.
The amending regulation was published in the Official Journal on 24 July 2026 and entered into force on 27 July 2026. It defers the two high-risk tracks and moves the national sandbox deadline. It adds one prohibition.
It does not touch the prohibitions already in force, the AI literacy duty, the general-purpose AI chapter or the Article 50 transparency obligations. English coverage written before July 2026 often still gives the old dates, so check the publication date of whatever you are reading, including this page.
One thing worth stating plainly, because it is the most common misreading: a system is high-risk because of what it does, not because of the sector it sits in. Annex III is organised by use case. A recruitment tool that screens candidates falls under it; one that only drafts job adverts does not.
No. Only the high-risk obligations moved. Regulation (EU) 2026/1744 shifted Annex III to 2 December 2027 and Annex I to 2 August 2028. The prohibitions and AI literacy have applied since 2 February 2025, GPAI rules since 2 August 2025, and the Article 50 transparency duties since 2 August 2026.
Yes. Article 50 took effect on 2 August 2026 and the Digital Omnibus did not defer it. One narrow exception: generative systems already on the market before that date have until 2 December 2026 to add the machine-readable marking of synthetic content required by Article 50(2). Systems placed on the market after that date must comply immediately.
Three things, and none of them wait for the high-risk deadline. Know which AI systems you use and what each one does, because classification depends on function rather than sector. Take demonstrable AI literacy measures, which have been required since February 2025. Check whether Article 50 applies to any system that talks to people or generates content.
Yes, where the output is used in the EU. Article 2 covers providers placing systems on the EU market regardless of where they are established, and providers and deployers outside the EU whose system output is used inside it.
The rest of this site is in Dutch. These two pages are not.
Reading Dutch? The full version of this page is at /kennisbank/ai-act-deadlines.
