DeepMind's Neel Nanda: Losing Monitorable Chain-of-Thought Would Be a Safety Tragedy
NeelNanda5 · x · 2026-09-03
DeepMind interpretability researcher Neel Nanda forcefully pushed back on the "concerningly common" take that keeping Chain of Thought monitorable doesn't matter because interpretability will save us — calling it "total bullshit" and arguing CoT is our best current safety tool.
On the OpenAI looped transformer discourse, he says a model with 2x GPT-4's effective layers isn't scary, but the slippery slope toward many loops would be super bad.
More from Safety
- Center for AI Safety Publishes Full Syllabus for Intro to ML Safety Course — ArtificialOther · 2026-10-21
- Ben Goertzel outlines a general recipe for superintelligence 'basic drives': what it improves vs. protects — bengoertzel · 2026-10-21
- America.gov launches AI portal unifying 29,000 government sites with fraud reporting — soleio · 2026-10-21
- EvilWAF: open-source transport-layer MITM proxy that bypasses modern WAF behavioral detection — godfrey_rono · 2026-10-21
- 2022 interview resurfaced: AI safety researcher says competence, not consciousness, is the risk — davidmanheim · 2026-10-21
- Domingos: GDPR purpose-limitation makes LLM pretraining illegal in Europe — pmddomingos · 2026-10-21