US Senate Discusses AI Regulation Bill That Would Require Companies to Prove They Took Reasonable Precautions

The US Senate is discussing a bill to regulate AI. Reuters reports the bill would require AI companies to prove they have taken reasonable steps to prevent harm from their tools. Lawmakers are considering giving the US Commerce Secretary power to demand evidence that AI developers have met a "duty of care," and to send government auditors to test company products directly.

Last week, Anthropic researcher Jacob Kocson announced his resignation, saying part of the reason was that people building AI genuinely believe the technology could kill everyone before the end of the 2020s. His statement fueled fears about AI risks. Several AI leaders then called for slowing down development, and global AI stocks fell on the 14th.

People familiar with the talks say negotiators are also considering letting federal courts handle cases where the US government believes certain AI models are unsafe and wants to block their release. Companies could challenge the government's decision in federal court. The federal court provision would also bar states from enforcing their own laws on certain AI model risks. The bill is still in negotiation, and even if Congress passes it, it has little chance of becoming law.