Anthropic Launches Free AI Scanner for Open-Source Projects

Anthropic has launched OSS Scanner, a free, opt-in service that uses artificial intelligence to scan selected open-source projects for potential security vulnerabilities and send findings to maintainers. The reports are generated by models without human review or triage before delivery, the company says, and may contain errors or incorrect severity assessments. Core maintainers can apply through GitHub, with Anthropic assessing projects individually. The company says it will consider criteria similar to Google’s OSS-Fuzz, including a project’s importance to critical infrastructure and security. Applying does not guarantee acceptance. On Friday, Nethermind, ZEUS and VirtEngine were among projects reported to have submitted requests; none of those cited requests had been merged at the time of publication. Anthropic says enrolled projects will receive periodic scans at no cost, using its strongest models, including Claude Mythos. Reports are intended to help maintainers investigate potential issues and may include a reproducer, an explanation and a suggested patch. The company says a report may also identify where a vulnerability was introduced when possible.