Krux
February 23, 2026
Anthropic's Claude Now Hunts Software Vulnerabilities Autonomously
Published: February 23, 2026 at 12:28 AM
Updated: February 23, 2026 at 12:28 AM
100-word summary
Anthropic launched Claude Code Security, an AI vulnerability scanner now in limited research preview for Enterprise and Team customers. Unlike traditional tools, it reasons about code, traces data flows, uses multi-stage verification, and suggests patches rather than auto-applying them. Security teams can scan pull requests for flaws, and developers review AI-generated patch recommendations before pushing fixes. During internal testing, Opus 4.6 identified over 500 zero-day vulnerabilities in production open-source code. Security engineers benefit most from automated detection across large codebases, though findings still require human approval to avoid false positives.
What happened
Anthropic launched Claude Code Security, an AI vulnerability scanner now in limited research preview for Enterprise and Team customers. Unlike traditional tools, it reasons about code, traces data flows, uses multi-stage verification, and suggests patches rather than auto-applying them. Security teams can scan pull requests for flaws, and developers review AI-generated patch recommendations before pushing fixes. During internal testing, Opus 4.6 identified over 500 zero-day vulnerabilities in production open-source code.
Why it matters
Security engineers benefit most from automated detection across large codebases, though findings still require human approval to avoid false positives.