Back to Devoured - May 25, 2026
DEVOURED
Anthropic prepares Mythos 1 for Claude Code and Claude Security

Anthropic prepares Mythos 1 for Claude Code and Claude Security

AI securityllmenterprise Testing Catalog
Anthropic is nearing a broader public release of Claude Mythos 1, its AI model specialized in cybersecurity, for Claude Code and Claude Security enterprise offerings.
What: Anthropic's Claude Mythos, previously restricted, is moving towards a general release as Mythos 1, with evidence appearing on Google Cloud and AWS. It's being prepared for use in Claude Code and Claude Security, which now includes a new dashboard for surfacing vulnerabilities. Project Glasswing, an AI cybersecurity initiative, has already found over 10,000 high- or critical-severity vulnerabilities using Mythos-grade models. Claude Opus 4.8 is also rumored for release soon.
Why it matters: This marks a strategic expansion for Anthropic into dedicated enterprise cybersecurity and code analysis, leveraging its advanced models to directly address critical business needs beyond general-purpose LLM applications, but also suggests the company is still cautious about widespread availability of such powerful tools.
Takeaway: If your organization uses Anthropic's enterprise offerings, expect enhanced code and security analysis capabilities with Mythos 1, potentially leading to faster vulnerability discovery.
Deep dive
  • Anthropic is preparing Claude Mythos for a broader release as Mythos 1.
  • Evidence of Mythos 1 has appeared with a "claude-mythos-1-preview" label for Claude Code and Claude Security.
  • Project Glasswing, using Mythos-grade models, has already identified over 10,000 high- or critical-severity vulnerabilities.
  • The Claude Security offering is receiving a new dashboard for discovered vulnerabilities, historical charts, and triage results.
  • Earlier statements suggested Mythos would remain restricted, making this a notable shift towards broader availability once safeguards are in place.
  • Claude Opus 4.8 is also rumored to be in internal evaluations with partners, potentially launching in the coming weeks.
Decoder
  • Project Glasswing: Anthropic's collaborative AI cybersecurity initiative focused on using AI models to discover software vulnerabilities.
Original article

Anthropic appears to be moving Claude Mythos closer to broader availability than its original guidance suggested. A recent Project Glasswing update notes that the model is now helping protect a wider range of organizations, including open-source projects, and adds that Mythos-grade models could reach the public once the right safeguards are in place.

Last month we launched Project Glasswing, our collaborative AI cybersecurity initiative. Since then, we and our partners have found more than ten thousand high- or critical-severity vulnerabilities in essential software.

— Anthropic (@AnthropicAI) May 22, 2026
And in the near future, once we’ve developed the far stronger safeguards we need, we look forward to making Mythos-class models available through a general release.

This marks a notable shift from the earlier framing, in which Anthropic stated that Mythos would remain restricted. Traces of the model have already surfaced on Google Cloud and AWS through vulnerability discovery programs, and signals now point to a product called Mythos 1, carrying a preview label ("claude-mythos-1-preview"), being prepared for Claude Code and Claude Security.

Some users were temporarily able to see the "Mythos 1" model in the UI. Besides that, new strings have been added to the source code recently:

"Access to the Claude Mythos model in Claude Code and Claude Security."

The Claude Security side of that rollout is getting structural work too. A new dashboard is being built that surfaces discovered vulnerabilities, seven-day and thirty-day historical charts, and deeper triage results. There are no indications yet that the product will move beyond enterprise customers, but the refresh brings it closer to parity with how rival security suites present scan history.

In parallel, Claude Opus 4.8 is rumored to be in the works for release, with select Anthropic partners already conducting internal evaluations. A launch in the coming weeks would fit the cadence set by Opus 4.7 in April and would slot neatly alongside the Mythos and security product moves.

Explore all 62 articles in today's digest

Fresh dev news, AI-digested daily. Never miss a trend at devoured.fyi