Unauthorized group has gained access to Anthropic’s exclusive cyber tool Mythos, report claims

Concerns are mounting after reports surfaced that an unauthorized group has accessed Mythos, Anthropic's recently unveiled cybersecurity tool. Mythos, designed specifically for enterprise security, has attracted attention due to its potential risks if misused. According to a report by Bloomberg, a clandestine online forum managed to breach access to Mythos via a third-party vendor. An Anthropic spokesperson confirmed that they are investigating claims of this unauthorized access, stating, "We’re investigating a report claiming unauthorized access to Claude Mythos Preview through one of our third-party vendor environments." However, the spokesperson also indicated that there is currently no evidence suggesting that Anthropic's internal systems were compromised. The group reportedly employed various tactics to gain access, including leveraging the credentials of an individual working for a contractor associated with Anthropic. Members of this forum are said to be part of a Discord channel focused on discussing unreleased AI technologies. Since gaining access, they have been experimenting with Mythos and provided evidence of their usage to Bloomberg, including screenshots and a live demo of the tool’s capabilities. Interestingly, the unauthorized access coincided with the tool's public announcement. The group purportedly deduced Mythos's online location based on Anthropic's previous model formats. While they claim their intention is not to misuse the tool for malicious purposes, the potential for such misuse remains a significant concern for Anthropic. Mythos was intended for a select group of vendors, including major corporations like Apple, as part of Project Glasswing, aimed at preventing its exploitation by malicious actors. With the unauthorized use of Mythos now a reality, Anthropic faces serious questions about the effectiveness of its security measures designed to protect corporate environments from cyber threats.