Read the AI Act like a diff.
Every verified change to the EU AI Act — what moved, who it affects, and when it takes effect. Cited to the official source. Then check, in three questions, whether it applies to you.
See which rules apply to you →or subscribe to the feed
In force applies now · Upcoming future effective date · Deferred deadline pushed later · Proposed no effective date yet
- In force
EU-level AI Act — Art. 53–55 · Ch. V: general-purpose AI (GPAI) value changed upcoming → in force — since 2 Aug 2025
Providers of general-purpose AI models (the foundation models other products build on) must keep technical documentation, publish a summary of the content used to train the model, and put a policy in place to respect EU copyright law. Models judged to carry systemic risk face extra testing, incident-reporting and cybersecurity duties.
- In force
EU-level AI Act — Art. 5: prohibition value changed upcoming → in force — since 2 Feb 2025
Eight categories of AI are banned outright in the EU — including social scoring, untargeted scraping of facial images to build recognition databases, emotion recognition in workplaces and schools, and systems that manipulate or exploit vulnerable people. Providers and deployers must not place these on the market or put them into use.
- Deferred
EU-level AI Act — Art. 6(2) · Annex III: high-risk value changed 2 Aug 2026 → 2 Dec 2027
The full obligations for high-risk AI in the Annex III use cases — employment, creditworthiness, education, biometrics, essential public and private services and more — cover risk management, data governance, technical documentation, logging, human oversight, accuracy and cybersecurity, plus a conformity assessment and EU-database registration before going to market. The deadline, originally 2 August 2026, was deferred to 2 December 2027 by the Digital Omnibus on AI (Regulation (EU) 2026/1744, in force since 27 July 2026).
- In force
EU-level AI Act — Art. 4: governance value changed upcoming → in force — since 2 Feb 2025
Any company that provides or uses AI must ensure the people operating it on their behalf have a sufficient level of AI literacy — enough understanding to run the systems responsibly and grasp their risks. It applies to every risk tier, not just high-risk AI.
- Deferred
EU-level AI Act — Art. 6(1) · Annex I: high-risk value changed 2 Aug 2027 → 2 Aug 2028
High-risk obligations for AI that is a safety component of products already covered by EU product-safety law (machinery, medical devices, toys, lifts, vehicles). The deadline, originally 2 August 2027, was deferred to 2 August 2028 by the Digital Omnibus on AI (Regulation (EU) 2026/1744, in force since 27 July 2026).
- In force
EU-level AI Act — Art. 99 · Ch. XII: governance value changed upcoming → in force — since 2 Aug 2025
The Act's governance framework and penalty regime take effect. Breaching the prohibited-AI rules can cost up to €35M or 7% of worldwide annual turnover; most other breaches up to €15M or 3%. National market-surveillance authorities and the EU AI Office begin oversight.