European Union Artificial Intelligence Act

Assembled from 32 blocks · 2 evidence · 30 related

1. Story
2. The first comprehensive statutory regime for AI systems, in force from 1 August 2024, structured by risk tier rather than by technology: prohibited practices, high-risk systems carrying conformity obligations, transparency duties, and a separate track for general-purpose models. Its interest for this Journey is what happened to its timetable. The obligations for high-risk systems were to apply from 2 August 2026; a Digital Omnibus proposal published on 19 November 2025 sought to defer them; the Council and Parliament agreed a text on 7 May 2026; and the amending act was published in the Official Journal on 24 July 2026 and entered into force on 27 July 2026, moving the Annex III stand-alone high-risk deadline to 2 December 2027 and the Annex I embedded-system deadline to 2 August 2028, while transparency duties and the AI literacy obligation kept their original dates. A regime designed for a technology whose capability was expected to be assessable found the assessment machinery — the harmonised standards it depends on — was not ready in time. That is a measurement problem expressed as law.
3. Knowledge
4. Machine learning in medical imaging
5. The difficulty of evaluating machine learning systems
6. European Union Artificial Intelligence Act
7. Connections
8. The difficulty of evaluating machine learning systems
9. Machine learning in medical imaging
10. AI standards bodies
11. The EU Artificial Intelligence Act enters into force
12. The Commission proposes deferring the high-risk obligations
13. The deferral becomes law
14. High-risk obligations under the AI Act become applicable
15. CASP — Critical Assessment of Structure Prediction
16. Construct validity in AI evaluation
17. Benchmark saturation
18. METR
19. European Union Artificial Intelligence Act
20. A human-parity claim in translation fails under document-level evaluation
21. A randomised trial finds experienced developers slower with AI assistance
22. An audit finds the most-cited preference leaderboard is optimised rather than merely measured
23. A systematic review of 445 benchmarks reports pervasive validity failures
24. The Commission proposes deferring the high-risk obligations
25. The AI Index records both fast benchmark movement and doubts about benchmarks
26. Deep learning
27. European Union Artificial Intelligence Act
28. Geoffrey Hinton
29. A randomised screening trial reports what AI-supported mammography actually changes
30. Evidence
31. The AI Act entered into force on 1 August 2024 with staged application. The Digital Omnibus on AI, published 19 November 2025, proposed deferring the high-risk compliance deadline from 2 August 2026; the Council and Parliament reached political agreement on 7 May 2026; the amending act was published in the Official Journal on 24 July 2026 and entered into force on 27 July 2026, deferring Annex III stand-alone high-risk obligations to 2 December 2027 and Annex I embedded-system obligations to 2 August 2028, while transparency rules and the AI literacy duty retained their original timeline. V55 verification basis: neither the Official Journal nor the Council page was fetched. The dates and the substance were carried in search results attributing them to the Council's May 2026 press release and to law-firm and specialist briefings on the Omnibus; the Official Journal citation number of the amending act was NOT retrieved and the regulation number given in this record's title should be confirmed by a curator before publication.
32. ISO/IEC JTC 1/SC 42 was created in 2017 with its inaugural meeting in April 2018 and is the primary international committee for AI standardisation; ISO/IEC 42001:2023, published in December 2023, is the first AI management-system standard and adopts the clause structure of other ISO management-system standards so that organisations can be certified against it; NIST published the voluntary AI Risk Management Framework (AI 100-1) in January 2023, organised around the Govern, Map, Measure and Manage functions; CEN-CENELEC established JTC 21 in June 2021 to draft European standards including the harmonised standards supporting the EU AI Act, and announced measures in October 2025 to accelerate that work. V55 verification basis: none of these pages was fetched and ISO/IEC 42001 is paywalled and was not consulted. Every date and description above was carried in search results attributing it to the named body's own pages or documents; the standard's contents are described from secondary summaries.