Anthropic, an AI research and safety company, has introduced a new tool titled OSS Scanner designed to assist open-source developers in identifying security vulnerabilities within their codebases. The service is offered free of charge to projects that choose to opt-in. According to the company, the scanner utilizes Anthropic’s most advanced AI models to perform periodic, thorough reviews of project repositories to detect potential security flaws.

A significant feature of the OSS Scanner is its reliance on automation. Unlike traditional security audits that often involve human cybersecurity experts to verify findings, Anthropic’s service provides outputs that are entirely model-generated. There is no human review or triage process involved in the reporting. While this approach allows for faster and more frequent scanning cycles, it also introduces the possibility of model-related errors or false positives that would typically be filtered out by human analysts. The initiative is positioned as a way to help smaller open-source projects improve their security posture by providing access to high-level diagnostic tools that might otherwise be resource-prohibitive. Developers using the service are expected to manage the findings independently, as the AI acts as a diagnostic assistant rather than a managed security service.