AI Math Proofs Slash Software Verification Costs From Decades To Weeks

Anthropic has deployed its Claude model to mathematically verify the correctness of existing software code, a milestone that transforms a historically labor intensive process. By leveraging the AI to generate vast amounts of proof code, the company aims to make AI generated software more trustworthy. This shift reduces the reliance on manual expert review for complex system validation. The effort marks a significant step toward automating the assurance of digital infrastructure reliability.

The traditional method of formal verification has been prohibitively expensive for most applications. For example, specialists required twenty person years to verify seL4, a small operating system kernel. Because of this high cost, the technique was largely reserved for extreme cases such as Airbus avionics software. Anthropic's Frontier Red Team, working with partner Theorem, handed this burden to Claude. Within weeks, the model generated approximately 180 million lines of mathematical proof code. It drew on test programs from 123 open source libraries in Google's OSS-Fuzz security project to perform this work.

Claude did not write new software but instead verified existing code by converting crash tests into mathematical problems. It analyzed compiled machine code and constructed extensive proofs that a separate program called Rocq checks automatically. Anthropic states the process requires almost no human reviewers. The researchers describe the proof generation as fully autonomous, even though humans built the underlying tools and formal foundations at Theorem. This autonomy allows for a scalable approach to checking complex code bases without the previous bottleneck of specialist availability.

During the verification process, Claude encountered hundreds of bugs and was permitted to fix them. Many of these defects likely escaped conventional testing because they appear only under narrow conditions. These conditions include unusually large input files or specific sequences of accesses. In one week of verifying a widely used PDF reader, the team found as many security bugs as are usually reported for that software in a full year. In another case, Claude identified a logical error that had gone unnoticed for more than a decade. Even when a complete proof fails, the attempt can still reveal real bugs.

The company is careful to note that formally verified does not mean error free. The proofs hold only under defined conditions, such as a particular processor architecture and single threaded programs. Inputs must also be smaller than 64 megabytes. Real world deployments of these libraries often run on different hardware and configurations, which limits the universal applicability of the current proofs. This distinction is crucial for developers planning to adopt these verification methods in production environments.

Cost remains another constraint for the technology. Individual proofs consume hundreds of millions of tokens, and Anthropic has not published total project costs. However, the company argues that if AI can verify its own code, developers could trust AI generated software more and spend less time reviewing it. A smaller test run indicated that proofs can be reused and efficiency is improving. Opus 5.5 solves half the proofs using about 40 percent fewer tokens than the previous version, Opus 5.

This project shifts the focus from whether AI can program to whether it can prove the code it writes is safe. Anthropic expects formal verification with AI to become as practical as standard testing. Theorem's roadmap already targets critical infrastructure components including glibc, OpenSSL, and cURL. This outlook suggests a future where mathematical certainty is a standard part of software development rather than a rare luxury reserved for high stakes systems.