Anthropic is making significant moves with its renowned AI, Mythos, which is specifically designed for cybersecurity. Recent indications strongly suggest its integration into Claude Code and Claude Security. The key evidence comes from references to Mythos 1, tagged as “claude-mythos-1-preview,” alongside updates to the Claude Security panel. Although a full public launch announcement hasn’t been made yet, these are clear signals that Anthropic is preparing for broader user accessibility.
This development is particularly noteworthy because Mythos wasn’t created as just another classic chatbot, text summarizer, or enhanced programming tool. Anthropic unveiled it as part of Project Glasswing, a cybersecurity initiative involving prominent companies and organizations such as AWS, Apple, Google, Microsoft, NVIDIA, and the Linux Foundation. The project’s objective is to identify critical vulnerabilities in real-world software before malicious actors can exploit them. The progress observed recently is impressive.
According to TestingCatalog, Anthropic is reportedly preparing Mythos 1 for Claude Code and Claude Security. The most direct clue is a string found in the code mentioning “access to the Claude Mythos model in Claude Code and Claude Security.” Furthermore, some users have briefly seen the Mythos 1 model within the interface, labeled as a preview, which aligns with a yet-to-be-officially announced rollout.
The sensitive nature of Mythos lies in its capabilities. Anthropic has previously explained that this type of model can identify high-severity vulnerabilities and, in certain scenarios, assist in their exploitation.
This is precisely why the company had restricted its access to specific partners and for defensive use cases, rather than releasing it as a public feature within Claude. The shift now is that Anthropic itself has hinted that, once stronger safeguards are in place, it anticipates bringing Mythos-class models to a general release.
Claude Security also appears to be undergoing a significant update. The new dashboard is expected to display discovered vulnerabilities, historical 7- and 30-day graphs, and more detailed triage results. This aligns with a tool designed for security teams, enterprises, and developers who need to review flaws, prioritize risks, and understand which issues require immediate attention.
Claude Code is poised to be the other logical recipient of this advancement from Anthropic. If Mythos is integrated there, the company could transform its assisted programming environment into a much more potent tool for code review, bug detection, and identifying security issues before they reach production. However, there is currently no confirmation that Mythos 1 will be available to all users from day one, nor whether it will be released outside the Enterprise environment.
What we have are clear signs of preparation, but not a finalized official release. Mythos 1 is appearing in internal signals, Claude Security is being bolstered, and Anthropic is already discussing general availability for Mythos models once sufficient security measures are established. For an AI capable of finding real-world vulnerabilities, the line between a defensive tool and a potential risk will be the most closely monitored aspect of this entire deployment.
