Leaked documents show Anthropic’s new model poses unprecedented cybersecurity risks

According to reports linked to Fortune, leaked internal documents from Anthropic have revealed the existence of a new generation of AI models codenamed “Claude Mythos.”

However, the bigger headline is that the new models are reportedly already in testing phase despite being believed to pose “unprecedented cybersecurity risks.”

Anthropic’s data leak

Fortune recently released a report containing leaked documents from the artificial intelligence company, Anthropic. The documents reveal details about a new generation of AI models being developed, reportedly named “Claude Mythos.”

The leak occurred due to a human error in Anthropic’s content management system (CMS) configuration. Roy Paz from LayerX Security and Alexandre Pauwels from the University of Cambridge explained that Anthropic left nearly 3,000 unpublished assets, including images, PDFs, audio files, and draft blog posts, in a publicly accessible and searchable data lake.

The material was discoverable online before Fortune notified Anthropic, after which the company removed public access.

The documents viewed by Fortune claim that Claude Mythos is already in the testing phase. More importantly, they state that Anthropic believes this new model “poses unprecedented cybersecurity risks.”

Anthropic stated in a draft blog that the system is “currently far ahead of any other AI model in cyber capabilities” and warned that “it presages an upcoming wave of models that can exploit vulnerabilities in ways that far outpace the efforts of defenders.”

Due to the potential risks, Anthropic planned a cautious rollout strategy that would prioritize early access for cybersecurity defense organizations, giving defenders a head start in strengthening codebases against AI-driven exploits.

Anthropic has previously reported that a Chinese state-sponsored group used Claude Code to infiltrate about 30 organizations, including tech companies, financial institutions, and government agencies, so the concern is not unwarranted.

The leaked documents also revealed details of an invite-only CEO summit planned at an 18th-century manor in the English countryside, where Anthropic CEO Dario Amodei was set to host European business leaders to discuss AI adoption and showcase unreleased Claude model capabilities.

Elon Musk’s continued battle with Anthropic

Once the news broke on X and garnered significant attention, Elon Musk, the owner of X and the head of xAI, a direct competitor to Anthropic, did not miss the opportunity to comment on the news.

Musk commented, “Seriously troubling,” and quickly gained tens of thousands of views and thousands of likes.

Musk has a pattern of commenting on negative news regarding competitors with whom he disagrees. Anthropic was founded by former OpenAI employees, and Musk has been openly critical of both OpenAI and the general AI industry’s approach to safety and commercialization.

Meanwhile, Musk’s own AI company, xAI, recently launched a new paid subscription tier called “SuperGrok Lite”. The subscription costs $10, and limits have been placed on free users of Grok to push them towards buying it or any other of the three subscription plans that Grok offers.

They include SuperGrok, which costs $30 a month, SuperGrok Heavy, which costs $300 a month, and Grok Business, which also costs $30 per month.

Source