Claude Security now scans continuously and tracks fixes (claude.com)

Claude Security has launched an enhanced version that continuously scans and tracks vulnerabilities in software repositories without requiring users to initiate scans manually. Since its public beta debut in April, the tool has helped thousands of organizations discover over a million vulnerabilities. The updated version now automates monitoring, keeping pace with code changes and automatically marking findings as fixed when deployment occurs. This proactive approach significantly reduces the risk window and enhances security response by integrating a knowledge base for user-defined security policies and threat models tailored to specific environments. This continuous monitoring is significant for the AI/ML community because it leverages the capabilities of Claude's family of models, including Claude Mythos, to provide users with real-time insights into potential vulnerabilities, prioritizing fixes based on contextual threat levels. Additionally, improvements such as an API for integration into existing workflows and new role-based access controls empower organizations to streamline their security operations. The ability to analyze severity and exploit paths for each vulnerability allows teams to tackle the most critical issues first, promoting a more effective and efficient development process.