A leaked Anthropic AI model shook the cybersecurity sector, causing significant stock declines among major corporations. Initially seen as a potential breakthrough, this incident now prompts critical questions: Can organizations trust AI with their most sensitive digital assets? Or does this event merely underscore the necessity for expert protection?

Mint reported in a

leaked draft blog post

that a new AI model class called Capybara has been introduced. The draft suggested that Capybara outperformed Anthropic’s flagship model, Claude Opus 4.6, in “software coding, academic reasoning, and cybersecurity-related tasks.” Additionally, training on Claude Mythos—a model described as the most advanced by Anthropic—has also been completed.

Why Did It Leak?

Anthropic attributed the leak to “human error,” but this explanation might not allay concerns about the company’s data security capabilities. Some analysts suspect alternative motivations.

“The leak of Capybara is unfortunate, yet I can’t help but wonder if it was intentionally left in an accessible data lake to highlight some of the emerging cyber risks posed by continually evolving AI platforms,” stated

Tracy Goldberg

, Director of Cybersecurity at Javelin Strategy & Research. “Nonetheless, the model is still in testing, with Anthropic clearly stating that it is aware of bugs and risks needing to be addressed. This is why Anthropic has only soft-launched Capybara.”

The Looming Threat of AI

Anthropic also warned about the cybersecurity risks associated with these models, underscoring the escalating arms race between defenders and cybercriminals. They cautioned that Capybara could be the first in a series of models capable of identifying and exploiting vulnerabilities faster than security teams can respond. In other words, criminals could leverage the model to fuel a new generation of AI-driven cybersecurity threats.

The financial markets reacted quickly, with shares of CrowdStrike, Datadog, and Zscaler dropping more than 10% in early trading.

“The sharp decline in tech stocks following the Capybara leak underscores a lack of understanding among investors about AI,” Goldberg noted. “We know these models will continue to adapt at a pace faster than industry security measures can respond. This is why governance around AI is so critical.”