- A leak this week revealed Anthropic’s most powerful AI model yet, the “Claude Mythos.”
- The new model’s advanced Cybersecurity capabilities triggered a sell-off in major cybersecurity stocks.
- The leak originated from draft materials stored in an unsecured, publicly accessible data cache.
- The model introduces a new tier above Opus, with a second version internally called “Capybara.”
A leaked draft post this week prematurely revealed Anthropic‘s development of Claude Mythos, described internally as its most capable AI model. The existence of the advanced system was first reported by Fortune after published files were found in a publicly accessible cache.
According to an archived page, the company stated Mythos is “the most powerful AI model we’ve ever developed” and scored dramatically higher than its Opus models in coding and cybersecurity. Anthropic attributed the data exposure to human error in configuring its content management system tools.
Consequently, cybersecurity stocks experienced a sharp decline following the reports. Shares of firms like Palo Alto Networks and CrowdStrike dropped around 7% and 6.4%, respectively, according to Yahoo Finance.
However, Anthropic acknowledged the model’s significant risks, noting it “presages an upcoming wave of models that can exploit vulnerabilities.” The company plans a cautious, limited early-access release focused on defense organizations.
This market reaction mirrors the sell-off triggered by Anthropic‘s Claude Cowork reveal in February, which erased billions in market value. The leaked documents also revealed a second version of the model, internally named “Capybara,” positioned above the current Opus tier.
✅ Follow BITNEWSBOT on Telegram, Facebook, LinkedIn, X.com, and Google News for instant updates.
