Software stocks and bitcoin moved lower after reports about leaked Anthropic materials describing a more capable Claude model and warning about cybersecurity risks.
What happened
Fortune reported that Anthropic has started testing a new AI model that the company called “the most capable we’ve built to date.” According to the report, the model is being tested with a small group of early-access customers while Anthropic evaluates its behavior and risks.
Details surfaced after internal materials were accidentally exposed in a publicly accessible data store, Fortune said. The exposed files reportedly included draft announcements and internal content tied to Anthropic’s blog.
Key details from the leak report
One draft blog post referred to the model as Claude Mythos. The document said the system could pose serious cybersecurity risks because it could identify and exploit software vulnerabilities.
The leaked materials also suggested Anthropic is developing a new model tier called Capybara, described as larger and more intelligent than Opus, which the company currently presents as its most advanced model tier.
Market moves cited in the source
CoinDesk said Palo Alto Networks, CrowdStrike, and Fortinet were down 4% to 6% on the news, while the iShares Expanded Tech-Software Sector ETF fell 2.5%.
CoinDesk also said the overnight leak likely contributed to bitcoin falling back toward $66,000 after trading near $70,000 earlier.