EU AI Act Faces Collapse Amid Foundation Model Disputes
Germany, France, and Italy are blocking strict rules for foundation models in favor of industry self-regulation, threatening to derail the landmark EU AI Act.
Ongoing trilogue negotiations over the EU AI Act face a major roadblock as Germany, France, and Italy reject comprehensive regulation for foundation models. Instead of adhering to a previously agreed graduated approach that applies stricter rules to more powerful AI systems, these countries push for self-regulation through a voluntary code of conduct to protect their domestic AI companies. This unexpected reversal threatens to jeopardize the entire legislative effort.
In response to this pushback, the European Commission proposes a compromise text that keeps the tiered regulatory structure but significantly weakens the overall rules. This watered-down proposal now serves as the foundation for continued deliberations, though it faces strong expected resistance from the European Parliament, which advocates for tighter oversight of general-purpose AI systems like ChatGPT.
Time is running out for lawmakers to reach a consensus before the final trilogue in December, as a change in Council Presidency and upcoming European Parliament elections in 2024 loom large. Failure to resolve these foundational disagreements risks considerable delays or the complete collapse of the act, undermining the EU's position as a global pioneer in artificial intelligence regulation.