Formal verification's AI moment: Anthropic proves 239 entry points with 180M-line Rocq proofs
vinodg · x · 2026-11-18
A SIGPLAN essay on why formal verification is having a moment, with AI as both problem and solution: agents find bugs in critical software while also completing verifications once needing decades of human effort. Key facts: Theorem's $134M nixpkgs roadmap, Harmonic ($295M) and &Truth ($70M) raises; Anthropic & Theorem proved 239 entry points across 123 open-source libraries never crash using 180M lines of machine-checked Rocq; Google verified Signal security properties in 6.5M lines of Lean. The author urges making verification cheap and expressive enough to survive the hype.
More from Venture
- Paid AEO tools aren't more accurate — start with OpenAI's free first-party reporting — 5le · 2026-11-18
- Prompt research beats keyword research for content ideas, argues SEO veteran — 5le · 2026-11-18
- Nvidia's $100bn+ AI investment web under scrutiny as $500bn finance platform lands zero deals — GaryMarcus · 2026-11-18
- Ex-SEC official flags Nvidia's $52B Singapore revenue disclosure shift — MatthewChang · 2026-11-18
- Microdramas made $11B last year; AI cuts $200K series budgets 100x — moplee1990 · 2026-11-18
- Felicis leads $7.5M round for Unsiloed AI to make enterprise docs agent-ready — TechCrunch AI · 2026-11-18