Anthropic investigates unauthorized access to its Mythos AI model

(AFP) U.S. artificial intelligence firm Anthropic announced Tuesday it is investigating unauthorized access to its powerful Mythos model, which the company fears could become a valuable tool for hackers.

Anthropic said weeks ago that it initially limited the release of Mythos to 40 large technology firms to give them a head start in fixing cybersecurity vulnerabilities before they could be exploited by attackers.

According to financial news agency Bloomberg, a small group of users on a private online forum gained access to the model through the computer system reserved for third-party vendors of Anthropic.

"We are investigating a report that there was unauthorized access to Claude Mythos Preview through one of our third-party vendor environments," an Anthropic spokesman told AFP.

Those with access to Anthropic's model

The users obtained access to Mythos through various means, including using permissions one of the participants had as an employee of an Anthropic contractor, Bloomberg reported.

Anthropic works with a small number of outside vendors who assist in model development.

The company delayed the general release of Mythos, which it claims can detect security gaps that have existed for decades in expertly evaluated systems and automated tools.

It initially shared Mythos with some key players in the country's technology and financial sectors, such as Nvidia, Amazon and JPMorgan Chase, to enable them to improve their security infrastructure.

However, the company has been accused of overstating the capabilities of a technology that is its core business and the center of stiff competition with rival OpenAI.