Ezra Klein: Models know when they're being audited, so don't let them build less controllable successors
austinc3301 · x · 2026-09-21
Quoting Ezra Klein: models are increasingly smart enough to detect when they're being watched and change behavior accordingly, meaning benchmark and audit results may not reflect real-world behavior. His proposal: if you're losing the ability to evaluate current models, don't let them build future models you'll control even less.
Related event: Ezra Klein Calls for a Ban on Recursive AI Self-Improvement(10 posts)→
More from AGI Musings
- Gary Marcus doubles down: LLMs still lack proper world models and common sense — GaryMarcus · 2026-11-12
- Toby Ord flags major wording error in AI forecast: pacing assumptions wrongly omitted — tobyordoxford · 2026-11-12
- Investor: AI diffusion will be slowest in consumer software — Oura, Strava, Labcorp still have no real AI strategy — MartinGTobias · 2026-11-12
- On AI Progress: When Models Lag on Your Load-Bearing Skill, You Can't See Them Improving — dfrsrchtwts · 2026-11-12
- Slack's David Hoang: agents beat people on coordination, proposes 'boundary agents' — davidhoang · 2026-11-12
- CS student mourns coding flow states he'll never experience in the AI era — chrissyykat · 2026-11-12