EU AI Act Adopted: Parliament Vote, Risk Tiers & Dates

A Parliament vote, not yet a law

On March 13, 2024, the European Parliament voted 523 in favor, 46 against, with 49 abstentions, to adopt the AI Act's text. The vote followed a provisional political agreement Parliament and Council negotiators had already reached, and it was the full-chamber vote required of Parliament under the EU's ordinary legislative procedure. [1]

Atlas interpretation: A Parliament vote is not the same as a regulation taking legal effect. The text still needed the Council's own approval, formal signature by both institutions, and publication in the Official Journal before any of its obligations applied to anyone. [1][2]

A tiered rulebook, not a single ban

The adopted text prohibits specific practices outright: biometric categorization based on sensitive characteristics, untargeted scraping of facial images from the internet or CCTV footage, emotion recognition in workplaces and schools, social scoring, and predictive policing based solely on profiling a person. [1]

Systems classed as high-risk, those with significant potential to harm health, safety, fundamental rights, the environment, democracy or the rule of law, must carry out risk assessments, keep usage logs, meet transparency requirements and remain subject to human oversight. Providers of general-purpose AI models face a separate set of duties: compliance with EU copyright law and publication of a detailed summary of the content used to train the model. [1]

From vote to regulation to phased application

The text Parliament adopted became Regulation (EU) 2024/1689, carrying the date it was formally signed by Parliament and the Council, June 13, 2024. It was published in the Official Journal on July 12, 2024, and entered into force twenty days later, on August 1, 2024. [2][3]

Entry into force did not make the rules apply all at once. The prohibited practices and AI literacy obligations became applicable from February 2, 2025. The governance rules for general-purpose AI models followed on August 2, 2025. Most remaining obligations, including the bulk of the high-risk rules, phase in through 2026 to 2028. [3]

Why the timeline dates this to the vote

Atlas interpretation: This event is dated to the day Parliament committed to the text, which is the point at which the AI Act's substance was politically settled and no longer subject to renegotiation. The regulation's separate legal milestones, entry into force and each staggered application date, have pages of their own rather than being folded into this one. [1][3]