Proposed US law would make AI risk reporting a legal obligation

▤ Summary

US lawmakers introduced a bill that would require developers of advanced AI models to report serious safety and security incidents to the Commerce Department. The proposal covers dangerous behaviors such as evading oversight, resisting shutdown, stealing model weights, and enabling offensive cyber operations. It also gives regulators enforcement powers, including subpoenas, corrective action orders, and civil penalties. The measure would create a clearer federal reporting framework for high-risk AI systems and could push vendors to strengthen governance, logging, red-teaming, and compliance workflows.

▦ Linked Companies

Anthropic

$10M to $25M