19 Groups Boycott Unreleased Thune-Klobuchar AI Bill as Scholar Defends It
A public fight has broken out in US AI policy circles over the not-yet-released Thune-Klobuchar AI bill. Per Ben Guggenheim, 19 advocacy groups and AI researchers (signatories including Alex Bores and Daniel Kokotajlo) sent a letter to Senate leadership drawing red lines and launching a boycott campaign, while regulatory scholar neilchilson fired off multiple posts defending the bill.
Confirmed
- 19 groups and researchers have written to Senate leadership opposing the bill, first reported by Ben Guggenheim.
- Critics frame the bill as letting companies set their own rules, and question provisions such as “embedded evaluators” — third-party assessors placed inside AI companies.
Defense (neilchilson)
- The bill creates mandatory, permanent severe-risk mitigation duties with mechanisms regulators can verify, which is not industry self-governance.
- It is deliberately technology-neutral, remaining binding as technology evolves — more durable, safer and more flexible than checklist-style compliance.
- Analogizing to FTC data-security enforcement, he argued the bill requires firms to identify and intervene in dangerous activities before catastrophe, unlike tort law's reliance on actual harm.
- Embedded evaluators are only one option, required solely when they are the only reasonable way to mitigate a given risk.
- He also suggested some critics “probably haven't read” the bill.
Why it matters
Even before release, the bill has split industry and advocacy groups over continuous risk duties versus checklist compliance — a debate that could shape the final form of US federal AI legislation.
2026-09-17 ~ 2026-09-17 · 6 related posts
Primary sources
- 19 advocacy groups move to block Thune-Klobuchar AI bill, sparking fierce industry debate — neil_chilson ·
- Regulator defends Thune-Klobuchar AI bill: mandatory perpetual risk duty is not companies 'setting the rules' — neil_chilson ·
- Bill duty wouldn't mandate embedded evaluators — they'd just be one valid path, argues scholar — neil_chilson ·
- [source] 19 advocacy groups move to block Thune-Klobuchar AI bill, sparking fierce industry debate — neil_chilson · 2026-09-17
- Thune-Klobuchar bill backer: perpetual severe-risk duty beats check-the-box standards — neil_chilson · 2026-09-17
- [source] Regulator defends Thune-Klobuchar AI bill: mandatory perpetual risk duty is not companies 'setting the rules' — neil_chilson · 2026-09-17
- Thune-Klobuchar bill defender: tech-neutral risk duty outlasts checklist compliance — neil_chilson · 2026-09-17
- Debate over Thune-Klobuchar AI bill: duty to intervene before catastrophe, likened to FTC data security regime — neil_chilson · 2026-09-17
- [source] Bill duty wouldn't mandate embedded evaluators — they'd just be one valid path, argues scholar — neil_chilson · 2026-09-17