European Union officials reached a landmark deal Friday on the world’s most sweeping bill to regulate artificial intelligence, cementing the bloc’s role as the de facto global tech regulator, as governments scramble to address the risks created by rapid advances in AI systems.
Details about the compromise were not immediately available following three days of marathon talks, dragged out by disagreements between E.U. legislators and powerful member states, including France, Germany and Italy.
The E.U. legislation establishes a hierarchy of regulations, putting the most stringent limits on systems that policymakers have deemed the riskiest. Late in the negotiations, France and other member states objected to provisions in the bill regulating the models that underpin generative AI products like the popular chatbot ChatGPT, which generates human-like speech by crunching data scraped from across the internet. Law enforcement’s use of facial recognition also emerged as a critical flash point in the talks.