AI Research Atlas

Claude Code Security

Anthropic · 20 February 2026

Claude Code Security scans codebases for vulnerabilities and proposes patches; Anthropic says Opus 4.6 found 500+ long-undetected bugs in open-source software.

A reasoning-based vulnerability scanner rather than a pattern matcher. It traces business logic and access control flaws, verifies findings in multiple stages and requires human approval for patches. Limited research preview for Enterprise and Team, and open-source maintainers can apply for free expedited access.

Date
Friday, 20 February 2026
Lab
Anthropic
Kind
product
Access
research preview

Figures

MeasureValueMeasured by
Vulnerabilities found in production open-source code500+
using Claude Opus 4.6; company claim, bugs undetected for decades
company

The 500+ count is Anthropic's claim (detailed in its Opus 4.6 zero-days post, 2026-02-05) and was not independently verified here. A broader Claude Security product entered public beta on 2026-04-30 (see anthropic-claude-security-public-beta).

Sources

  1. www.anthropic.com/news/claude-code-security

This record was checked against its sources on 6 October 2026. How we check

Related