Anthropic prepares for early launch of Mythos AI model
What's the story
Anthropic's highly anticipated AI model, Mythos, could be launching sooner than expected. The speculation comes from a report by Testing Catalog, which found references to "claude-mythos-1-preview" in Anthropic's source code and Claude Security interface. This development hints at the possibility of the AI start-up bringing its advanced AI model closer to public access through tools like Claude Code and Claude Security.
Cloud presence
Mythos spotted on Google Cloud and AWS
The Testing Catalog report also noted the presence of the Mythos model on Google Cloud and AWS through a vulnerability discovery program. Some users even got a glimpse of the 'Mythos 1' model in Claude. This further strengthens the theory that Anthropic is preparing for an early launch of its advanced AI technology.
Dashboard update
New Claude security dashboard in the works
Along with the possible launch of Mythos, Anthropic is also working on a new Claude Security dashboard. The updated interface will show discovered vulnerabilities, historical charts for seven and 30 days, and detailed triage results. As for Mythos itself, it's Anthropic's most powerful model to date. The company has previously described it as the first model capable of surpassing all but the most skilled humans at finding and exploiting vulnerabilities.
Security concerns
Mythos has already discovered thousands of high-severity vulnerabilities
Since its introduction, Mythos has found thousands of high-severity vulnerabilities across operating systems, browsers, and open-source software projects. Anthropic had initially withheld the model over concerns about potential misuse. However, the company has now said that once it develops necessary safeguards against misuse, it intends to make Mythos-class models available via a general release.