Did Claude Mythos find thousands of zero-day exploits in OS?

Analysis Reasoning

The post claims Anthropic announced and withheld a model called Claude Mythos Preview for finding thousands of zero-day exploits—this is MISLEADING. A leak exposed internal documents on Claude Mythos (Capibara) with such capabilities, but Anthropic made no public announcement, per the only available evidence from a YouTube report on the accidental exposure.

Cited Sources

• 01
  Secureworld

  https://www.secureworld.io/industry-news/anthropic-claude-mythos-finds-exploits-zero-days

• 02
  Elisity

  https://www.elisity.com/blog/claude-mythos-ai-vulnerability-discovery-microsegmentation-unpatchable-devices

• 03
  Helpnetsecurity

  https://www.helpnetsecurity.com/2026/04/08/anthropic-claude-mythos-preview-identify-vulnerabilities/

• 04
  Thehackernews

  https://thehackernews.com/2026/04/anthropics-claude-mythos-finds.html

• 05
  Penligent

  https://www.penligent.ai/hackinglabs/claude-mythos-preview-and-the-new-zero-day-era/

• 06
  Forums Freebsd

  https://forums.freebsd.org/threads/ai-finds-thousands-of-zero-day-exploits-including-in-freebsd.102288/

• 07
  Youtube

  https://www.youtube.com/watch?v=6NP_N9FRb8M

• 08
  Tomshardware

  https://www.tomshardware.com/tech-industry/artificial-intelligence/anthropics-claude-mythos-isnt-a-sentient-super-hacker-its-a-sales-pitch-claims-of-thousands-of-severe-zero-days-rely-on-just-198-manual-reviews