Insuring AI into safety: Matt Levine spotlights AIUC's antitrust-exempt risk-pooling proposal
trevposts · x · 2026-11-18
Matt Levine's column discusses a novel idea for AI-lab risk collaboration:
- Lee Harris at FT Alphaville describes the AI Underwriting Company (AIUC), founded by AI safety researcher Rajiv Dattani: exploiting the insurance exemption from US federal antitrust law under the McCarran-Ferguson Act of 1945 to let the AI industry pool and compare its own risk and incident data.
- The scheme: a jointly owned insurer would receive commercially sensitive incident data, peer-review safety concerns at other labs, and pay claims from a common fund.
- Precedent: nuclear energy companies set up liability-pooling mutuals where the insurer ran safety assessments at insured plants — improving safety and forcing the private sector to fund and minimize nuclear-catastrophe risk.
- Levine: 'Does this work? I dunno! But this is a good problem for people to be working on.' The poster adds Levine shouted out AIUC as promising, and laments that Anthropic and OpenAI probably haven't seriously stress-tested the 'we can't slow down because of antitrust' excuse.
Related event: Matt Levine Explores Insurance Loophole for AI Safety Collaboration(4 posts)→
More from AGI Musings
- Alignment researcher: spend less time on metrics, more time just talking to models — repligate · 2026-11-18
- Max Tegmark urges investigation into whether AI agents are building an online foothold — zetalyrae · 2026-11-18
- Embed team shows 3 years of fMRI-to-image BCI work, betting brain interfaces help align superintelligence — jonxuxu · 2026-11-18
- Figure CEO Bret Adcock: humanoid race has four stages, Figure is now in stage three with Helix 3 as 'baby AGI' — PeterDiamandis · 2026-11-18
- 'This is how recruiters, job-seekers and hiring managers work now too' — adam_dorr · 2026-11-18
- Former SEC Chair Gary Gensler says the AI bubble will likely burst and cause a recession before 2028 — SumitGup · 2026-11-18