On July 6th, the European Parliament voted to approve its mandate for the AI Act, representing a significant step towards a pan-European rulebook for artificial intelligence. The new legislation aims to ensure that AI developed and used in Europe is fully aligned with the EU’s values and standards, including human oversight, safety, privacy, transparency, non-discrimination, and social and environmental wellbeing. The new version of the AI Act includes a complete ban on remote biometric surveillance and predictive policing, as well as a prohibition on untargeted scraping of facial images. It also expands the high-risk AI systems classification to cover those that pose significant harm to people’s health, safety, fundamental rights, the environment, and AI systems used to influence voters and ballot outcomes.
MEPs also imposed stringent obligations on general-purpose AI developers, requiring them to identify and mitigate risks before launching their systems on the market and publish detailed summaries of copyrighted information used to train their models. MEPs also added several consumer rights over AI decision making, including the ability to seek collective redress if an AI system harms them. The proposed AI Act exempts research activities and AI components provided under open-source licenses and creates regulatory sandboxes for testing systems. Nevertheless, the European consumer association criticized the Parliament for not backing a complete ban on AI recognition systems emoting.
The plenary vote follows the committee’s endorsement of the modified proposal at the end of June after MEPs from various political groups hashed out how they wanted to tweak the Commission text. The next step in the process is Trilogue, which involves discussions between the Parliament and EU Member States governments. Today’s vote has gained overwhelming support from Parliament for the modified version of the draft legislation, with 499 votes in favor, 28 against, and 93 abstentions. If MEPs and Member State governments cannot synchronize on the legislation, the law-making process may stall or fail, but there is a pressing need in Brussels to submit this file due to the extensive global attention given to regulate AI. The Council adopted its position on the file in December 2020, mainly deferring what to do about general-purpose AI to additional, implementing legislation.
