Why Mythos could be launching sooner than you think? Leaked code exposes preparations for Claude Mythos 1

Anthropic’s Mythos AI model, whose mere announcement sent shockwaves across cybersecurity and banking sectors, could be gearing up for release if recent leaks are to be believed. As per a report by Testing Catalog, references to a model called “claude-mythos-1-preview” were recently spotted inside Anthropic’s source code and Claude Security interface. This suggests that the AI startup could finally be aiming to bring its powerful AI model closer to general access via tools like Claude Code and Claude Security.

The report also notes that traces of the model have surfaced on Google Cloud and AWS via the vulnerability discovery programme. It adds that some users were able to see the ‘Mythos 1’ model in Claude.

Testing Catalog also reports that Anthropic is building a revamped Claude Security dashboard that shows discovered vulnerabilities, seven-day and thirty-day historical charts, and deeper triage results.

What is Claude Mythos?

Claude Mythos is Anthropic’s most powerful model yet, which it first unveiled last month and called the first model to reach a level where it can surpass all but the “most skilled humans” at finding and exploiting vulnerabilities.

Anthropic had refused to release the model publicly over concerns about misuse.

“Models with advanced offensive cybersecurity capabilities present obvious dual-use risks,” the company said at the time, adding that it wanted to introduce the technology “gradually and with strong safeguards in place.”

The company also said that Claude Mythos Preview had already discovered thousands of high-severity vulnerabilities across operating systems, browsers and open-source software projects.

In a recent blog post, Anthropic revealed that since sharing the model with its partners in April, it has been used to find over 10,000 “high- or critical-severity vulnerabilities across the most systemically important software in the world”.

“Progress on software security used to be limited by how quickly we could find new vulnerabilities. Now it’s limited by how quickly we can verify, disclose, and patch the large numbers of vulnerabilities found by AI,” the company added.

While Anthropic recently stressed that Mythos-class models reduce the time and cost required to find and exploit vulnerabilities, the company also noted that once it develops the required safeguards, it plans to make Mythos-class models available via a general release.

“At present, no company, including Anthropic, has developed safeguards strong enough to prevent such models from being misused and potentially causing severe harm. That is why we have yet to release Mythos-class models to the public,” Anthropic noted.

“But it’s also why we began Project Glasswing: if a similarly capable model is released without such safeguards, it will soon become dramatically cheaper and easier for almost anyone in the world to exploit flawed software,” the company added.

The change in stance from Anthropic, along with the recent report, could be hinting at the new model arriving sooner than anticipated.