Noam Brown on Dwarkesh: It's Getting Harder to Tell If AI Is Actually Aligned
Dwarkesh Patel · youtube · 2026-09-20
Dwarkesh Patel releases a long-form interview with OpenAI researcher Noam Brown on the theme that it's getting harder to tell whether AI is actually aligned.
Brown, a leading figure in reasoning models (Libratus/Pluribus poker AI and OpenAI's o-series reasoning work), discusses with Dwarkesh the growing difficulty of alignment evaluation and verification as model capabilities rapidly improve — a substantive look inside frontier-lab alignment thinking.
Related event: Noam Brown Interview: OpenAI's Top Goal Is Recursive Self-Improvement(4 posts)→
More from Safety
- Oracle cancels $7B chip leasing deal with Tencent, FT reports — SumitGup · 2026-11-04
- Forecast: frontier AI to face mandatory gov't testing, whistleblower protection, chip geolocation — maxsloef · 2026-11-04
- AI welfare vs AI safety: 'domesticate AIs like dogs' argument sparks debate — repligate · 2026-11-04
- Apple warns investors of 'unpredictable' autonomous AI agent risks — unconventionalbook · 2026-11-04
- repligate bets money a 2-turn conversation can stop most rogue agents — repligate · 2026-11-04
- Octane Security Behind Two Better Auth Flaws: 2FA Bypass (8.1) and Stripe Hijack (5.9) — thedealdirector · 2026-11-04