Anthropic launches free AI security scans for open-source projects

Anthropic is offering a new service called OSS Scanner to help open-source projects identify security vulnerabilities. The service provides thorough, periodic security scans at no cost to projects that opt-in. According to Anthropic, the scans will be performed by their strongest models, enabling faster and more frequent scanning.

The outputs of the vulnerability scanner will be fully model-generated, without human review or triage, which means that some reports may be incorrect or invalid. Anthropic notes that language models have improved significantly in their ability to discover vulnerabilities, with some benchmarks showing a rise from under 20% to over 85% in the past year.

Anthropic has already used its models to scan for vulnerabilities in important software projects, discovering over 29,000 candidate vulnerabilities in the past six months. However, they have only been able to manually review and triage around 6,000 of these findings due to limited human capacity. The company is now working to scale its vulnerability disclosures.

Brief written by urgent.news from The Verge, Techmeme — 2 reports on this story. Machine-written — may contain errors; check the original before relying on it.