A handful of unauthorized users in a private Discord channel have been accessing Anthropic's Mythos model since the day the company announced it

Anthropic has a security problem. A small group of unauthorized users have been quietly accessing the company's new Mythos AI model — reportedly since the very day it was announced.

The breach was traced to a private Discord channel where a handful of individuals managed to get their hands on the technology without authorization. Bloomberg's Rachel Metz broke the story, citing sources familiar with the situation.

Anthropic has described Mythos as exceptionally powerful, which makes the unauthorized access all the more concerning. The company hasn't publicly detailed how the users gained entry or how long the exposure went undetected.

The incident raises serious questions about access controls at one of the world's most prominent AI labs. When you're building technology you yourself call dangerously capable, keeping the gates locked seems like table stakes.