113/69 Wednesday, February 25, 2026
Anthropic has introduced Claude Code Security, an AI-driven security service built on top of its Claude Code platform. The tool is designed to scan source code for security vulnerabilities and provide remediation guidance in the form of suggested patches for development teams to review. The service is currently available as a limited research preview for Enterprise and Team customers, and open-source project maintainers may also apply for access.
According to the company, Claude Code Security is engineered to detect complex and high-severity vulnerabilities that traditional rule-based scanning tools often miss. Instead of relying solely on predefined patterns, the system analyzes code structure and data flow, understands relationships between components, and identifies weaknesses across interconnected systems. Each detection undergoes a multi-stage verification process, where the model attempts to validate or refute its own findings to reduce false positives. Vulnerabilities are also severity-ranked and presented in a human-review dashboard to help teams prioritize remediation effectively.
Anthropic stated that the tool builds upon more than a year of research into the cybersecurity capabilities of its Claude models, including participation in Capture-the-Flag exercises and collaboration with the Pacific Northwest National Laboratory. Using the Claude Opus 4.6 model, the team reportedly identified over 500 previously undiscovered vulnerabilities in open-source projects. The company emphasized that AI-powered code analysis will play an increasingly important role in the future, while also warning that malicious actors may leverage AI to discover vulnerabilities more rapidly. As a result, organizations are encouraged to adopt defensive AI tools proactively to reduce the risk of advanced cyberattacks.
