Anthropic uses Claude to formally verify 100+ open-source libraries at assembly level
xennygrimmato_ · x · 2026-10-24
Anthropic and Theorem show LLMs make formal software verification feasible at scale: Claude proved security properties at the RISC-V assembly level across 100+ real open-source libraries, reportedly the first large-scale assembly-level effort on production software. Claude also found bugs that survived trillions of test runs. Formal methods expert Hillel Wayne notes the key innovation: instead of translating source code to a theorem language (where translation errors prove the wrong thing), they compile to assembly and then to a theorem language called LPS, sidestepping source-semantics divergence — though you still must trust the hardware, OS, OCaml, and the translation itself.
More from Safety
- Apple's AI instructions literally said 'Do not hallucinate' — BBC false alerts persisted anyway — GaryMarcus · 2026-10-24
- 'Black October' security findings: AI agents make exploiting old flaws cost pennies — LexSokolin · 2026-10-24
- Google says it might use your Gmail and Photos data to target ads — The Verge AI · 2026-10-24
- IBM X-Force's Vibe Fuzzing Talk: LLM-Guided Fuzzing with Graph Inputs and Program State — dyn___ · 2026-10-24
- Greece passes nationwide social media ban for children under 15 — Polymarket · 2026-10-24
- AI Finds Software Flaws Faster Than Fix: 45,207 CVEs Logged by Late July — rohanpaul_ai · 2026-10-24