Major AI labs agreed to share safety frameworks and publish model cards detailing capabilities and risks.
Policymakers are weighing a federal framework that could preempt a patchwork of state rules.
Advocates called the moves a start but pressed for binding accountability and independent oversight.
