Anthropic Launches Claude Security Public Beta – AI-Powered Code Vulnerability Scanning

Anthropic recently announced the launch of the Claude Security public beta for its Claude Enterprise customers. Previously known as Claude Code Security, this tool is built on Anthropic's latest general-purpose AI model, Claude Opus 4.7, and is designed to automatically scan codebases to detect and fix software vulnerabilities. According to reports, Claude Team and Max users will also gain access shortly. Users can use the service directly via the Claude.ai sidebar or by visiting claude.ai/security.

Claude Security offers scheduled scanning and targeted scanning capabilities, simplifies integration with audit systems, and improves tracking of classified findings. Anthropic states that Claude reasoning through code like a security researcher during scans, rather than relying on traditional rule-matching patterns. For each vulnerability discovered, Claude provides detailed explanations including confidence level, severity rating, potential impact, and reproduction steps, helping teams quickly assess and remediate issues.

Currently, Claude Security has partnered with multiple security vendors including CrowdStrike, Microsoft Security, Palo Alto Networks, SentinelOne, TrendAI, and Wiz. Just a few weeks ago, Anthropic also launched Project Glasswing and its frontier model Claude Mythos Preview. As AI-driven security scanning tools move toward commercial adoption, enterprises can expect greater automation and deeper analysis in code security reviews. However, seamless integration with traditional security workflows remains a key factor for successful implementation.