Unauthorized Group Accesses Anthropic’s Exclusive Cyber Tool Mythos, Report Says

Reports indicate that a group of unauthorized individuals has accessed Mythos, the cybersecurity solution recently unveiled by Anthropic.

The capabilities of Mythos have generated significant attention, as it is an AI tool designed for enterprise security that could potentially be misused for hacking, according to the company’s warnings. Bloomberg reported that members of an unidentified “private online forum” have successfully accessed the tool via a third-party vendor.

“We’re currently looking into a report of unauthorized access to the Claude Mythos Preview through one of our third-party vendor environments,” an Anthropic representative shared with TechCrunch. The company stated that, as of now, there is no evidence that this unauthorized activity has negatively affected Anthropic’s systems.

The unauthorized group attempted various methods to access the model, including leveraging credentials from an individual who was interviewed by Bloomberg. This individual is reportedly employed by a third-party contractor collaborating with Anthropic, according to the news outlet.

Members of this group reportedly belong to a Discord channel focused on gathering information about unreleased AI models. They have been utilizing Mythos regularly since obtaining access and provided evidence to Bloomberg through screenshots and a live demonstration of the tool.

According to Bloomberg, the group allegedly gained access to the tool on the same day it was announced to the public, making an “educated guess” about its online location based on information about how Anthropic has organized other models. A source noted that the group is “interested in experimenting with new models, not causing destruction with them.”

Mythos was distributed to a select range of vendors, including major companies like Apple, as part of an initiative called Project Glasswing. This limited rollout was aimed at preventing misuse of the tool. Anthropic has cautioned that such a tool could be wielded as a weapon against corporate security, rather than enhancing it.

If these reports are accurate, the unauthorized usage of Mythos could pose significant challenges for Anthropic, which had aimed to mitigate concerns regarding enterprise security with the exclusive release.