Particle.news
Download on the App Store

Anthropic Launches Claude Code Security as Cybersecurity Stocks Slide

The research preview applies multi-stage verification with fixes requiring human approval.

Overview

  • Built into Claude Code, the new capability scans entire codebases, reasons about data flows, assigns severity and confidence scores, and proposes targeted patches for developer review.
  • Anthropic is limiting access to Enterprise and Team customers for a research preview and offering expedited access to open‑source maintainers while triaging findings through responsible disclosure.
  • The company reports its Opus 4.6 model enabled its Frontier Red Team to identify more than 500 previously undetected vulnerabilities in production open‑source projects.
  • Anthropic acknowledges dual‑use risks and says it is investing in safeguards and monitoring to prevent malicious use of the technology.
  • Following the announcement, shares of JFrog (around 24%), CrowdStrike (about 8%), Okta (over 9%), GitLab (over 8%), Cloudflare (about 8%), Zscaler (roughly 5%), and others declined, with market commentators noting the tool targets code auditing rather than many vendors’ core businesses.