A collective of unverified individuals has allegedly infiltrated Mythos, the cybersecurity application recently unveiled by Anthropic. Significant attention has been drawn to Mythos and its claimed capabilities—an artificial intelligence system engineered for corporate security that, if misappropriated, could transform into a powerful instrument for cyber intrusion, as stated by the company.

EMBED_PLACEHOLDER_0

Reports now indicate that a "private online forum," whose participants remain anonymous, successfully obtained entry to the tool via a third-party supplier. Anthropic has stated that, to date, its investigation has uncovered no indication that this alleged unauthorized activity affected any of the company's internal systems.

The unauthorized collective employed multiple tactics to breach the model, leveraging "access" available to an individual interviewed for the report. That individual is presently an employee at a third-party contractor engaged by Anthropic.

Members of this group participate in a Discord channel dedicated to uncovering details about AI models that have not yet been publicly released. Since obtaining entry, the group has been utilizing Mythos consistently and furnished evidence through screenshots and a real-time software demonstration.

It is reported that the group managed to access the tool on the identical day of its public announcement, having "made an educated guess about the model’s online location based on knowledge about the format Anthropic has used for other models."

According to the source, the group is primarily "interested in playing around with new models, not wreaking havoc with them."

Mythos was distributed to a curated set of vendors, which included prominent entities such as Apple, under an initiative named Project Glasswing. This restricted release was intended to prevent the tool's exploitation by malicious parties.

Anthropic has cautioned that the application could be repurposed to attack corporate security defenses rather than strengthen them. If verified, this unauthorized access to Mythos may pose a serious challenge for Anthropic, which had initiated the exclusive release to address and mitigate enterprise security concerns.