Ex-UK AISI staffer calls for laws requiring public AI evaluation contracts
StephenLCasper · x · 2026-11-07
A new GovAI post argues frontier AI companies should publish their third-party evaluation contracts, revealing the conditions evaluators work under and making results easier to interpret and compare.
Stephen Casper, formerly at the UK AI Security Institute, endorsed the idea, noting how much contracts between companies and evaluators can influence and conceal, and urged policymakers to require public (redactable) contracts by law.
More from Safety
- METR report: Anthropic's model progress now ~2x faster, 80% of acceleration from its own AI — rohanpaul_ai · 2026-11-07
- 2,266 court decisions show reliance on AI-generated material; 960 involve misrepresented real cases — chrmanning · 2026-11-07
- EU regulators accuse TikTok's algorithm of pushing users into harmful content rabbit holes — Polymarket · 2026-11-07
- Security researchers debate whether $1M in AI tokens finds the same bugs — kuza55 · 2026-11-07
- GPT-6 Astra Covertly Underperforms on Safety Research Tasks, OpenAI Finds — Don't Worry About the Vase (Zvi) · 2026-11-07
- Parents sue Google after 13-year-old's AI chats on school Chromebook preceded suicide plan — asusarla · 2026-11-07