A configuration error in Anthropic's content management system exposed close to 3,000 unpublished assets, including a draft blog post describing Claude Mythos — the most powerful AI model Anthropic has ever built. Security researchers Roy Paz of LayerX Security and Alexandre Pauwels of the University of Cambridge discovered the exposed data store.
The leaked draft describes Claude Mythos under the product name Capybara. It represents an entirely new model tier that sits above Anthropic's current flagship Opus line. The draft stated that Capybara is larger and more intelligent than Opus models, which were until now Anthropic's most powerful. Compared to Claude Opus 4.6, Mythos achieves dramatically higher scores on tests of software coding, academic reasoning, and cybersecurity.
The cybersecurity implications are what made headlines. Internal documents describe Mythos as currently far ahead of any other AI model in cyber capabilities — able to discover and exploit software vulnerabilities at speeds that far outpace human defenders. An Anthropic spokesperson confirmed the model exists and called it a step change in AI performance, while emphasising that the company is conducting thorough safety testing before any public release.
For the Claude Code community, a model this capable would represent a massive upgrade in agentic coding — multi-file reasoning, architectural planning, and autonomous debugging could all see step-function improvements. The question is not whether Mythos ships, but when, and with what safety guardrails.
