Anthropic's Secret "Claude Mythos" AI Model Leaked -- And It Could Change Everything

The AI world just got its biggest surprise of 2026. Anthropic accidentally left details of its most powerful unreleased model sitting in a publicly accessible data store -- and what got leaked has the entire tech industry talking.

What Is Claude Mythos?

A draft blog post stored in an unsecured, publicly searchable data cache revealed the new model is called Claude Mythos. Anthropic described it as "by far the most powerful AI model we've ever developed."

The same leaked documents also referenced a new model tier called "Capybara" -- described as larger and more capable than Anthropic's existing Opus models, and more expensive to run. Capybara and Mythos appear to refer to the same underlying model.

After Fortune alerted Anthropic to the leak, the company confirmed what the documents revealed. An Anthropic spokesperson called the model "a step change" and confirmed it is currently being tested with a small group of early access customers. Threads

Claude Mythos at a Glance

Why Is Anthropic Being So Cautious?

The capabilities are remarkable -- but so are the risks. The leaked document states the model is "currently far ahead of any other AI model in cyber capabilities" and could enable hackers to exploit vulnerabilities at a scale that far outpaces defenders.

Because of this, Anthropic's rollout plan focuses first on cybersecurity organizations, giving defenders a head start before the model reaches wider audiences.

This isn't new territory for Anthropic. The company previously reported that a Chinese state-sponsored group had run a coordinated campaign using Claude Code to infiltrate roughly 30 organizations -- including tech companies, financial institutions, and government agencies -- before Anthropic detected and shut it down. Threads

Artificial intelligence is advancing faster than most regulatory and security frameworks can keep up with -- and Claude Mythos may be the clearest proof yet.

How Did the Leak Happen?

Anthropic attributed the exposure to "human error," explaining that its content management system publishes digital assets to public URLs by default unless a user manually changes the privacy setting. Threads Close to 3,000 unpublished assets were reportedly accessible before the company locked things down.

Want to stay updated on the latest AI model releases and tech developments? Check out our full AI coverage at TechnoSports for the latest news in tech and beyond.