Anthropic Offers Free AI Scans for Open Source

Anthropic launches OSS Scanner, a vulnerability detection service for open source projects, powered by its most advanced models, including Claude Mythos. The offering is free and periodic, but without human triage of reports, promising increased frequency and the risk of false positives, as the influx of AI-generated reports is already overwhelming some stakeholders.

Reports without human triage, faster but sometimes erroneous

The analyses provided by OSS Scanner are not subject to any human review or triage, which is intended to increase the speed and frequency of scans. In return, some reports may be incorrect or invalid. This automation comes at a time when open source projects are already struggling to manage the influx of AI-generated reports, a situation that notably affects Linus Torvalds and Google.

OSS Scanner: free, opt-in, and powered by Claude Mythos

Anthropic is launching OSS Scanner, a service designed for open source projects to identify security vulnerabilities. Teams that opt into this service receive free periodic and in-depth scans. The reports, entirely produced by models and without human review, rely on Anthropic's most powerful systems, including Claude Mythos, with the goal of providing the greatest possible defensive advantage.

AI has already helped identify major flaws, including "Copy Fail"

OSS Scanner is part of a broader trend of AI tools applied to bug detection. Similar systems have recently contributed to uncovering significant security vulnerabilities in open source software. Among them, the "Copy Fail" bug affected nearly all Linux distributions in May.