Updated
Updated · The Register · May 25
Anthropic to Release Mythos Models Publicly, Expanding Access to Governments and More Users
Updated
Updated · The Register · May 25

Anthropic to Release Mythos Models Publicly, Expanding Access to Governments and More Users

3 articles · Updated · The Register · May 25

Summary

  • Anthropic plans to open its Mythos-class AI models to the public, marking a broader rollout beyond earlier limited availability.
  • Government users are among those gaining expanded access, while Anthropic is still keeping its AI flaw-finder private as it works out safety guardrails.
  • The move follows signs that claude-mythos-1-preview was being readied for wider use in Claude Code and security products after earlier traces appeared on major cloud platforms.
  • Anthropic’s staged release suggests it is widening commercial and institutional use of Mythos while holding back higher-risk security capabilities until controls are in place.

Insights

Who ultimately decides when a cyber 'chaos option' like Mythos is safe enough for the public?
Can AI that autonomously hacks systems truly be made safe, or has Anthropic created an uncontrollable digital weapon?
As AI devalues bug discovery, how will the cybersecurity industry and job market be reshaped in the coming year?