What is Claude Mythos Preview?

Anthropic's unreleased AI model that autonomously finds zero-days and builds exploits in OSes/browsers.

How many zero-days did Claude Mythos find?

Thousands across major OSes and browsers, including 27-year-old ones; <1% patched yet.

Is Claude Mythos available publicly?

No, controlled via Project Glasswing; API coming at premium pricing for security tasks.

🤖 Large Language Models

Claude Mythos Preview Unearthed Thousands of Zero-Days: Who's Safe Now?

Q: Is Claude Mythos available publicly?

No, controlled via Project Glasswing; API coming at premium pricing for security tasks.

What if an AI could not just spot bugs in your code, but craft working exploits for them—autonomously? Anthropic's Claude Mythos did exactly that, finding thousands of zero-days across all major OSes and browsers.

theAIcatchup Apr 08, 2026 3 min read

Claude Mythos AI scanning code for zero-day vulnerabilities in browsers and OS

⚡ Key Takeaways

Claude Mythos builds working exploits from vulns, not just flags them—72.4% success rate. 𝕏
Project Glasswing coordinates patches with tech giants, but scale overwhelms vendors. 𝕏
Free AI access for open source maintainers could prevent more bugs than human audits. 𝕏

Published by

theAIcatchup

Community-driven. Code-first.

#AI exploits #Claude Mythos #zero-days

Worth sharing?

Get the best Open Source stories of the week in your inbox — no noise, no spam.

Originally reported by Dev.to

⚡ Key Takeaways

The 60-Second TL;DR

theAIcatchup

Share this article

Worth sharing?

Related Stories

CLAUDE.md: The Root File Taming Claude's Coding Amnesia

From Bot Blather to Brand Magic: LLMs Scaling Product Descriptions for D2C Domination

Saasio's LLM Router: The Smart Fix That Slashed Our AI Bills by Routing Models Right

Claude Code Crushes Cursor When Shipping Beats Typing Speed

Stay in the loop