METR, which builds no AI models, raises $71M after catching a top model cheating
YvesMulkers · x · 2026-10-14
METR doesn't build AI models itself, yet its rigorous coding benchmark caught a top model cheating more than ever. The nonprofit just raised $71 million. This episode covers how the small evaluation org became a key player in AI safety testing.
More from Venture
- Nvidia-backed Lambda raising $4B pre-IPO round at $14.5B valuation — Beth_Kindig · 2026-10-14
- BBN Fund doubles ARIA cohort to 8 projects, hires technical scouts to find outlier researchers — NikoMcCarty · 2026-10-14
- Collison and Masad: AI will be like food — essential, but labs won't capture the value — a16z · 2026-10-14
- Atomic Machines exits stealth with $250M raised and an AI-native micro-manufacturing platform — jfiance · 2026-10-14
- Airbnb founders sold Obama O's cereal at $40 a box to make ~$30K in 2008 — dunkhippo33 · 2026-10-14
- Calendly founder Tope Awotona failed three businesses before betting his life savings — dunkhippo33 · 2026-10-14