Claude Security Enters Public Beta, Scans Code for Vulnerabilities with Opus 4.7

Anthropic’s Claude Security scans codebases for vulnerabilities and generates patches using Opus 4.7, following Project Glasswing’s discovery of thousands of zero-day flaws across every major OS and browser.
artificial-intelligence
cyber-security
Author

Kabui, Charles

Published

2026-05-04

Keywords

claude-security, opus-4-7, project-glasswing, vulnerability-scanning, anthropic

Download as PDF

Anthropic launched Claude Security in public beta on April 30 for Enterprise customers. Powered by Opus 4.7, it scans repositories by tracing data flows across files and reasoning about how components interact, rather than matching known patterns. Each finding includes a confidence rating, severity assessment, and a generated patch that opens directly in Claude Code. A multi-stage validation pipeline reviews every result before it reaches an analyst, cutting false positives. Claude Security builds on Project Glasswing (April 7), a 12-company coalition including AWS, Apple, Google, Microsoft, CrowdStrike, and the Linux Foundation. Anthropic committed $100M in credits for Glasswing participants using Mythos Preview, an unreleased model that autonomously found thousands of zero-day vulnerabilities across every major operating system and browser.

Static analysis tools flag known patterns and stop there. Claude Security reads source code, understands component interactions across modules, and catches context-dependent flaws that pattern matching misses. Early users report going from scan to merged patch in a single sitting instead of days of back-and-forth. Security vendors CrowdStrike, Palo Alto Networks, SentinelOne, and Wiz are also embedding Opus 4.7 into their own platforms.

Mythos Preview found a 27-year-old vulnerability in OpenBSD and chained Linux kernel exploits for full privilege escalation, all without human guidance. When AI discovers flaws that survived decades of expert review and millions of automated tests, the window between discovery and exploitation compresses. The race is whether defenders deploy these tools faster than attackers gain comparable ones.

For more on this week’s AI developments, see AI News Mashup: Grok 4.3 Drops Prices, Claude Gets Security Mode, Pentagon Signs AI Deals.

Sources:

Disclaimer: For information only. Accuracy or completeness not guaranteed. Illegal use prohibited. Not professional advice or solicitation. Read more: /terms-of-service

Reuse

GNU GENERAL PUBLIC LICENSE v3.0(View License)

Citation

BibTeX citation:
@misc{kabui2026,
  author = {{Kabui, Charles}},
  title = {Claude {Security} {Enters} {Public} {Beta,} {Scans} {Code}
    for {Vulnerabilities} with {Opus} 4.7},
  date = {2026-05-04},
  url = {https://toknow.ai/posts/claude-security-public-beta-opus-47-project-glasswing/},
  langid = {en-GB}
}
For attribution, please cite this work as:
Kabui, Charles. 2026. “Claude Security Enters Public Beta, Scans Code for Vulnerabilities with Opus 4.7.” https://toknow.ai/posts/claude-security-public-beta-opus-47-project-glasswing/.