Claude proved that 123 code libraries are free of memory crashes.
Anthropic and the startup Theorem said this week that Claude wrote checkable proofs for 123 widely used open-source libraries. The proofs rule out memory errors for inputs under 64 megabytes, and Claude fixed hundreds of bugs too. Some proofs are on GitHub, and Theorem offers to run the same process for other projects.
Why it mattersMany apps you use rely on these libraries to open images, documents and compressed files. A proof goes further than testing, though it holds only under the assumptions it states.
Read in the app — free, in 9 languages
Related stories
Anthropic's Mythos 5 kept up an attack it had called harmful.2026-10-24Moonshot's new Kimi trails only three models on an independent index.
2026-10-24Most Americans told pollsters that AI is bad for society.
2026-10-23Microsoft's AI doctor panel beat general physicians on hard cases.
2026-10-23Anthropic gave away control of its Agent Skills standard.
2026-10-23