European Parliament Nears Vote on AI Act Compromise Amendments
The European Parliament reaches a provisional deal on the AI Act draft, setting up a committee vote on controversial provisions regarding biometric identification and generative AI.
The European Parliament reaches a provisional political deal on its draft report for the Artificial Intelligence Act. After months of tense technical and political meetings between members of the IMCO and LIBE committees, lawmakers agree on a compromise text. The committee will vote on these amendments in the coming weeks, though the text remains subject to minor technical adjustments before the final decision.
The most controversial points in the compromise focus on biometric identification technologies and the rise of generative AI systems like ChatGPT. The proposed provisions include a strict prohibition on real-time remote biometric identification systems in publicly accessible spaces. Additionally, the text bans systems that use biometric traits to categorize individuals based on sensitive attributes.
Regarding generative AI, Members of the European Parliament draw a clear distinction between general-purpose AI systems and foundation models, which face stricter requirements. Providers of generative AI systems must ensure their technology aligns with EU law and publish a summary of training data that is protected under copyright law. These rules aim to address the rapid expansion of advanced AI while protecting fundamental rights.