AI Reports · Anthropic · 2026-10-22 · Blog / commentary
LLMs are making formal software verification feasible at a much larger scale
Claude formally proved security properties at the RISC-V assembly level across more than 100 real-world open-source libraries, in a collaboration with Theorem. While writing the proofs, it found bugs that had gone undetected for years despite extensive testing. The authors argue that LLMs can now produce formal proofs at large scale, making formal verification a practical part of secure software development.