EU AI Act (Regulation (EU) 2024/1689)¶
Reference
Citation: European Union. "Regulation (EU) 2024/1689 (Artificial Intelligence Act)" (2024). Type: policy. Link: artificialintelligenceact.eu.
What it is¶
The European Union regulation on artificial intelligence. It sets a risk-based classification with obligations for providers and deployers of high-risk systems.
Amended in 2026 by the AI Omnibus (Regulation (EU) 2026/1744): the AI literacy duty softened from "ensure" to "support" (Art 4), a new legal basis added for processing special-category data to detect and correct bias (Art 4a), and high-risk compliance dates postponed to 2 December 2027 (Annex III) and 2 August 2028 (Annex I).
The transparency obligations in Article 50 apply from 2 August 2026: disclosure that a person is interacting with an AI system (50(1)), machine-readable marking of synthetic outputs so they are detectable as artificially generated (50(2)), notice to people exposed to emotion recognition or biometric categorisation (50(3)), and labelling of deepfakes and of AI-generated text published to inform the public on matters of public interest, unless it passed human review with a person holding editorial responsibility (50(4)). Systems already on the market before that date have until 2 December 2026 for the Article 50(2) marking duty alone. The Commission's reading of Article 50 is distilled at Commission Guidelines on Article 50 transparency; the voluntary code provided for by Article 50(7) is ref-transparency-cop-2026.
Two scope exclusions matter for science. Article 2(6) excludes AI systems and models, including their outputs, developed and put into service for the sole purpose of scientific research and development. Article 2(8) excludes research, testing, and development before a system is placed on the market or put into service, but not testing in real-world conditions. Most scientific use of agents therefore falls outside the Act, and the record's practices are not compliance requirements.
Role in the record¶
- Grounds BP03: Article 25 (a substantial modification can turn a deployer into a provider; framed for high-risk systems, threshold not yet fully defined).
- Grounds BP04: Articles 14 and 26 (human oversight; deployer duties).
- Grounds BP07: Article 50(2) and 50(4) (synthetic outputs marked machine-readably and detectable as AI-generated; AI-generated public-interest text labelled), in application from 2 August 2026.
- Grounds BP09: Article 14 (oversight must be effective and names automation bias as a hazard) and Article 50(4) (the disclosure duty falls away only where human review was substantive and a person holds editorial responsibility).
Atom-level for/against detail and quotes are in the provenance data (assets/provenance.yml), keyed by practice atom.