European Union. AI / automated decisions. Regulation (EU)

AI ActRegulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act)

Partially in force

§1Summary

Synopsis

A risk-based product-safety style regulation of AI. It bans certain practices, imposes conformity, data governance, logging, transparency and human oversight duties on high-risk systems, sets transparency duties for chatbots and generative content, and adds obligations for general-purpose AI model providers. Deployers also have duties, including fundamental rights impact assessments for some.

Key themes
  • prohibited practices
  • high-risk systems
  • conformity assessment
  • GPAI
  • transparency
  • human oversight
  • AI literacy
  • FRIA
Unique features

It is the first comprehensive horizontal AI law. It applies to providers and deployers whatever their location if output is used in the EU, and phases obligations in over 2025–2028.

SaaS relevance

SaaS that embeds AI features carries provider or deployer duties: transparency for chatbots and generated content, AI literacy, and full high-risk compliance if used in Annex III areas such as HR or credit.

§2Obligations

Extraterritorial
Yes
Breach notification
Serious incident reporting for high-risk systems to market surveillance authorities (Art 73)
DPO
No
Cross-border transfers
None
Maximum penalty
€35m or 7% worldwide turnover (prohibited practices)

§3Dates

Key dates
  • Adopted
  • In force
  • Prohibitions and AI literacy 2025-02-02; GPAI and governance 2025-08-02; general application and Art 50 transparency 2026-08-02. Amended by the Digital Omnibus on AI, Regulation (EU) 2026/1744 (in force 2026-07-27): Annex III high-risk obligations postponed to 2027-12-02; Annex I (product) high-risk to 2028-08-02; new prohibition on non-consensual intimate imagery/CSAM generation and a 2026-12-02 grace date for Art 50(2) for generative systems already on the market.

§4Text and authority

Text
Authority
European AI Office (GPAI); national market surveillance authorities; European AI Board

§5Where it applies

Also applies in

§6Sources and verification

Sources consulted
Verification

High confidence.

Regulation (EU) 2026/1744 was published in the OJ on 2026-07-24 (per secondary source). I did not open the EUR-Lex text of 2026/1744.