Publicly consequential AI evals should be testable by the public, says researcher

evijit · x · 2026-09-13

Continuing his critique, evijit argues that networking-based eval access is fine in principle, but becomes problematic when those evals are tied to international AI governance and frontier pacing. Things that affect the public should be testable by the public at large.

Related event: METR Independence Under Fire: AI Audit Ecosystem Accused of Revolving Doors and Regulatory Capture(31 posts)→

Original post →

More from Safety

• OpenAI blocks weapons details for teens, adds auto-ending chats fixated on violence — rohanpaul_ai · 2026-10-31
• First machine-checked security proof for end-to-end x86 process execution published — banteg · 2026-10-31
• Stealthy agent worms would route model calls through compromised machines — matthew_d_green · 2026-10-31
• Halloween joke nails data harvesting: full candy bar for your dad's Gmail access — Danhightower · 2026-10-31
• Professor runs Pangram on 50 pre-ChatGPT self-published novels: zero false positives — TuhinChakr · 2026-10-31
• Researcher disputes Berkeley law professors' brief over Pangram false-positive claims — TuhinChakr · 2026-10-31

Safety channel →