ANTHROPIC has launched Claude Security in public beta to its Claude Enterprise customers, expanding Claude’s offerings with a tool formerly known as Claude Code Security. Built on Claude Opus 4.7, it scans codebases to find and patch software vulnerabilities, offering scheduled and targeted scans, easier integration with audit systems, and improved tracking of triaged findings, with no API integration or custom agent build required.
A company statement says organizations using Claude can start scanning today, with access for Claude Team and Max customers coming soon. Claude Security can be accessed directly from the Claude[.]ai sidebar or at claude[.]ai/security, and it explains each finding in detail, including confidence, severity, impact, and reproduction, while also generating instructions for a targeted patch.
Since public beta, it has been tested by hundreds of organisations of all sizes, and new features include a multi-stage validation pipeline to reduce false positives, scan scheduling, directory-level targeting, and export options to CSV or Markdown, plus webhooks to Slack or Jira. OpenAI has meanwhile launched GPT‑5.4‑Cyber and expanded its Trusted Access for Cyber programme, underscoring a competitive AI security landscape.