The European Parliament has approved final changes to the AI Act that would impose stricter transparency and risk-review requirements on general-purpose AI systems. The vote marks a major step in the EU’s effort to set rules for foundation models, including obligations aimed at making these systems more accountable before they reach the public.

Under the revised framework, developers of large-scale AI models would face greater disclosure demands and would need to assess potential risks linked to safety, discrimination, and misuse. Lawmakers say the goal is to reduce harms while still allowing innovation in a fast-moving sector that has drawn concern from regulators, civil society groups, and technology companies.

The amendments are expected to take effect in stages, with enforcement beginning in 2027. That timeline gives companies time to adapt, but it also signals that the EU intends to keep pushing ahead with one of the world’s most ambitious regulatory regimes for artificial intelligence.

The vote adds to a growing global debate over how to govern powerful AI tools without stifling competition. For Europe, the latest move reinforces a regulatory approach centered on transparency, oversight, and public protection.