Anthropic Launches Free AI Vulnerability Scanning Service OSS Scanner to Help Open-Source Software Prevent Backdoors

Anthropic has announced the launch of OSS Scanner, a free AI-powered vulnerability scanning service for open-source projects. The service uses Anthropic’s strongest AI model, Claude Mythos, to scan open-source code for deep-layer logical vulnerabilities that traditional static analysis tools may miss.

The service is designed to provide faster and more frequent scanning, allowing open-source communities to receive potential security alerts in real-time. However, this approach also means that all vulnerability reports will be automatically generated by the model without human review or classification.

Anthropic’s move is not purely altruistic, as the company’s own cloud infrastructure and AI foundations rely heavily on open-source code maintained by volunteers. The company is aware of the potential consequences of vulnerabilities in these underlying codes, as seen in recent high-profile incidents like the XZ Utils backdoor event.