Report Claims Illegal Access to Anthropic’s Mythos AI Model Under Probe

(MENAFN) According to reports, a group of unauthorized users has reportedly gained access to Anthropic’s AI model “Mythos,” a system described as capable of advanced cybersecurity-related functions, including identifying and exploiting vulnerabilities across major operating systems and web browsers.

Bloomberg reported that the group accessed the model through a private online forum on the same day Anthropic announced plans to provide limited testing access to selected companies. The report, citing sources, said the users have continued interacting with the model, although not for cybersecurity exploitation purposes.

Evidence shared with the publication reportedly included screenshots and a live demonstration of the model in use.

Anthropic has stated that Mythos is restricted to a small number of approved software partners specifically to allow controlled testing and strengthening of systems against potential cyber threats. The company’s safeguards are intended to limit exposure due to the model’s described capabilities.

The report also suggested that the users may have gained access through a combination of methods, including a third-party contractor and tools commonly used in cybersecurity research.

In response, an Anthropic spokesperson said the company is investigating the claims and examining whether access occurred through a third-party vendor environment. The company added that there is currently no evidence indicating that the issue extends beyond that environment or affects Anthropic’s core systems.

Sources cited in the report indicated that the group’s activity appears exploratory rather than malicious, with no evidence that cybersecurity-related prompts were used on the model.

The incident highlights ongoing concerns around controlled access to advanced AI systems and the risks associated with third-party infrastructure in AI deployment environments.

MENAFN22042026000045017281ID1111015791