Claude proves 123 code libraries work correctly in weeks

Anthropic has used its Claude model to mathematically prove that 123 existing open-source code libraries function correctly. The company states that the AI generated approximately 180 million lines of proof code in a matter of weeks, a task that traditionally requires extensive manual labor. This effort represents a shift in how software reliability is confirmed, moving away from standard testing toward formal verification.

Formal verification involves proving that software runs without error under all circumstances. The process is historically expensive and time-consuming. For example, verifying seL4, a small operating system kernel, took specialists 20 person-years. Consequently, the technique has largely been reserved for high-stakes environments such as Airbus avionics software, where errors can be catastrophic.

Anthropic's Frontier Red Team collaborated with Theorem to automate this process. Claude analyzed compiled machine code and converted crash tests into mathematical problems to construct proofs. A separate program called Rocq automatically checks these proofs, requiring minimal human review. The researchers described the proof generation as fully autonomous, though they noted that the underlying tools and formal foundations were built by humans at Theorem.

The project uncovered hundreds of bugs in the existing libraries. Many of these errors likely escaped conventional testing because they only appear under narrow conditions, such as unusually large input files or specific access sequences. In one instance, the team found as many security bugs in a widely used PDF reader in a single week as are usually reported for that software in a full year. In another case, Claude identified a logical error that had remained unnoticed for more than a decade.

Anthropic emphasizes that formal verification does not guarantee the code is error-free in every scenario. The proofs hold only under defined conditions, including a specific processor architecture, single-threaded programs, and inputs smaller than 64 MB. Real-world deployments often run on different hardware and configurations, which limits the scope of the guarantee.

Cost remains a constraint, as individual proofs consume hundreds of millions of tokens. Anthropic has not published the total project costs but argues that AI verification could reduce the time developers spend reviewing code. A smaller test run showed that proofs can be reused and that efficiency is improving, with Opus 5.5 solving half the proofs using about 40 percent fewer tokens than Opus 5.

Theorem's roadmap targets critical infrastructure libraries, including glibc, OpenSSL, and cURL. Anthropic expects formal verification with AI to become as practical as standard testing. The project shifts the focus from whether AI can program to whether it can prove the code it writes is safe.