The most dangerous model of human AI has fallen into the wrong hands

Anthropic’s Mythos AI model, a powerful cybersecurity tool that the company said could be dangerous in the wrong hands, was accessed by a “small group of unauthorized users.” Bloomberg Reports. An unnamed member of the group, identified only as a “third-party contractor for Anthropic,” told the publication that members of a private online forum entered Mythos via a combination of tactics, using contractor access and “commonly used online espionage tools.”

Claude Mythos Preview is a new general-purpose model capable of identifying and exploiting vulnerabilities “in every major operating system and every major web browser when directed to do so by the user,” according to Anthropic. Official access to the model is limited to a few companies through Glasswing Project Initiativeincluding Nvidia, Google, Amazon Web Services, Apple, and Microsoft. Governments They also look to technology. Anthropic currently has no plans to release the model publicly due to concerns that it could be used as a weapon.

“We are investigating a report alleging unauthorized access to Claude Mythos Preview through one of our third-party vendor environments,” an Anthropic spokesperson said in a statement. Bloomberg. Anthropic currently has no evidence that unauthorized access is affecting the company’s systems or going beyond the third-party vendor’s environment.

The model was reportedly accessed illicitly on April 7, the same day Anthropic announced it would release Mythos to a limited number of companies for testing. However, the group that gained the unauthorized access has not been publicly identified Bloomberg It reports that its members are part of a Discord channel looking for information about unreleased AI models.

The group arrived at the Mythos using knowledge of other Anthropic model formats obtained from more recent Mercure data breach To make an “educated guess” about their website. Members have been using Mythos regularly since gaining access – offering screenshots and a live demo of the model as proof Bloomberg – Although not for cybersecurity purposes in an attempt to avoid detection by Anthropic. The group also gained access to other unreleased Anthropic AI models, according to Bloomberg.