The European Parliament adopts its AI Act position

The European Parliament voted 499 to 28, with 93 abstentions, to adopt its negotiating position on the AI Act, the version of the text MEPs would carry into three-way talks (“trilogue”) with the Council of the EU and the European Commission. The original 2021 Commission proposal predated ChatGPT; the version Parliament adopted did not.

MEPs had added a category the Commission’s draft lacked: obligations specific to foundation models and generative AI. Providers of such systems would have to assess and mitigate risks before release, register their models in an EU database, disclose when content was AI-generated, and publish “detailed summaries” of the copyrighted material used in training — a provision aimed squarely at systems like ChatGPT. Parliament also widened the list of banned uses to include real-time biometric identification in public spaces, emotion-recognition in policing and workplaces, and predictive policing based on profiling, and added social media recommender systems reaching more than 45 million EU users to the high-risk tier alongside election-influencing systems.

The vote did not enact anything. It fixed Parliament’s opening position for negotiation with the Council, which held a different draft, and with the Commission; reconciling the three would take another six months of closed-door bargaining before political agreement was reached in December 2023. Parliament’s final adoption of that reconciled text did not come until March 2024, with the Council’s formal sign-off two months after that.

The June 2023 vote mattered because it set the terms of the fight rather than because it changed the law: it was the moment generative AI, absent from the original proposal, became a named subject of what would become the first comprehensive AI statute passed by a major regulatory bloc.