Anthropic says it partnered with Theorem to use Claude to formally prove security properties at the assembly level across 100+ real-world open-source libraries (Anthropic)
Introduction — Even mature, heavily tested software contains bugs, and LLMs are increasingly adept at finding them.
Anthropic has announced a partnership with Theorem, utilizing Claude AI to formally verify security properties at the assembly level across over 100 real-world open-source libraries. This marks the first large-scale effort to apply AI to formal verification at such a granular level. Claude not only proved these security properties but also uncovered bugs that had gone undetected for years, despite extensive testing.
As AI models continue to advance, formal verification is expected to become a crucial aspect of secure software development, even for well-tested software like the Linux kernel, which has recently seen over 1,000 vulnerabilities. Formal verification involves translating code into a mathematical model and then proving that it adheres to a defined set of security rules.
While this process is complex and requires significant expertise, LLMs like Claude are making it more accessible. The collaboration demonstrates that frontier models can autonomously generate correctness proofs for real-world software, providing a roadmap toward ensuring bug-free code. Anthropic and Theorem are publishing a subset of these proofs and offering a Claude skill to help developers begin their own verification processes.
Written by urgent.news from Techmeme's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
