A breach has occurred involving Mythos, Anthropic’s newly launched cybersecurity AI, as unauthorized users have exploited vulnerabilities to gain access. Reports indicate that members of a private online forum have accessed Mythos via a third-party vendor, raising concerns over the tool’s potential misuse.
Mythos, designed to enhance enterprise security, could become a dangerous weapon if it falls into the wrong hands. Anthropic confirmed they are investigating claims of this unauthorized access but asserted that so far, no adverse effects on their systems have been uncovered. The breach allegedly happened on the same day Mythos was publicly launched.
Sources suggest the group used tactics involving access rights held by an employee of a third-party contractor connected to Anthropic. These individuals are part of a Discord channel focused on obtaining information about unreleased AI models. They have reportedly used Mythos extensively since gaining access and have even shared screenshots and conducted live demonstrations of the software with a news outlet.
Bloomberg reported that the unauthorized group speculated about the location of Mythos online based on their understanding of Anthropic’s previous model formats. While they are curious about experimenting with the tool, they claimed they do not intend to cause chaos with it.
Mythos’s release was part of Project Glasswing, intended for a select group of vendors, including major companies like Apple. The limited rollout aimed to prevent misuse by malicious actors, reinforcing Anthropic’s commitment to enterprise security. However, if the unauthorized use persists, it could jeopardise Anthropic’s efforts to protect businesses against cyber threats, highlighting the delicate balance between innovation and security in the AI landscape.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
