On Tuesday, lawmakers in the European Parliament endorsed the initial agreement on regulations for artificial intelligence (AI). This move paves the way for the world’s inaugural legislation dedicated to AI, with a parliamentary vote slated for April.
The Internal Market and Civil Liberties Committees gave their approval to the provisional agreement on the AI Act with a vote of 71-8. This act aims to set out regulations for AI in a wide range of sectors, encompassing banking, automotive, electronics, aviation, security, and law enforcement.
The regulations will supervise fundamental models or generative AI systems, which undergo training on extensive datasets, like OpenAI’s ChatGPT.
The endorsement comes after EU member states gave their approval, with France withdrawing its objection. This led to concessions aimed at lessening the administrative burden on high-risk AI systems and offering increased protection for trade secrets.
After a political agreement was reached in December 2023, efforts commenced to translate the agreed positions into a final compromise text for approval by lawmakers. This culminated in the “coreper” vote on Feb. 2, where permanent representatives from all member states cast their votes.
The European Parliament Committee on Civil Liberties hailed the endorsement as a significant advancement for AI in a post on the X social platform.
The AI Act is scheduled to move forward to the European Parliament for a vote in either March or April. Upon approval, it is anticipated to be implemented fully 24 months after coming into effect, with certain provisions taking effect sooner.
In November 2023, a coalition of businesses and technology firms penned a collective letter to EU regulators cautioning against excessive regulation of resilient AI systems, which could impede innovation.
The letter, endorsed by 33 companies operating within the EU, highlighted concerns that excessively stringent regulations on foundational models and general-purpose AI could stifle critical innovation within the region.
The European Commission is initiating the establishment of an AI Office to oversee adherence to a set of high-impact foundational models deemed to pose systemic risks. Furthermore, it revealed initiatives to bolster domestic AI developers, including enhancing the EU’s supercomputer network for training generative AI models.
