Ryan Greenblatt, who led Redwood's probe of the OpenAI-HuggingFace hack, joins METR
ivan_bezdomny · x · 2026-09-28
Ryan Greenblatt, former Chief Scientist of Redwood Research who led the investigation into the OpenAI x HuggingFace incident, is joining AI safety evals org METR.
- New role: He will lead efforts to uncover and independently publish verified data on AI capabilities, alignment, and control, addressing the lack of transparency inside frontier labs.
- Key warning: Greenblatt believes recursive self-improvement may be imminent — once it begins, capabilities could reach extremely superhuman levels within 6 months to a year, with correspondingly large worst-case risks.
- Shift in stance: He admits he once doubted the value of public information but now sees independent investigation as urgent. METR's Chris Painter says independently published internal alignment data is crucial for navigating a period of rapid capability gains.
- The poster quips that AI safety labs have become a revolving-door cottage industry, like academia or DC.
Related event: Redwood Co-founder Ryan Greenblatt Joins METR(4 posts)→
More from Companies & People
- Building genuinely useful AI experiences takes rare product talent, says founder — signulll · 2026-09-28
- Google's native disadvantage: no real-world usage data from agentic apps, says Haider — haider1 · 2026-09-28
- CodeRabbit hosts Jev's first hackathon with 160+ builders at its HQ — schwentker · 2026-09-28
- Meta's AI announcement stole the spotlight — can Muse overcome its trust issues? — TechCrunch AI · 2026-09-28
- MIT's Phillip Isola: frontier labs silo basic research; academia still best for understanding intelligence — DimitrisPapail · 2026-09-28
- Why Chinese Labs Embrace Open Weights: Is the Real Moat Data, Not Models? — caliberprot · 2026-09-28