Rep. Liccardo proposes liability safe harbor for enterprises using CAISSI-certified safest AI labs

ShakeelHashim · x · 2026-11-18

Rep. Liccardo published a proposal for AI regulation that works through revenues: CAISSI would designate the lab doing the best job on AI safety, and businesses using that lab's models would get a safe harbor from liability for certain model-caused harms, incentivizing labs to invest in safety — which he calls a "Manhattan project for AI safety."

Commentator CharlieBull0ck analyzes the theory of change and doubts it: enterprise customers would buy safer models mainly for an affirmative defense in court, but labs are already adequately incentivized to build guardrails. Feedback is solicited on his Substack.

Related event: US Lawmaker Proposes Revenue-Based Incentives to Drive AI Safety(2 posts)→

Original post →

More from Safety

Safety channel →