Anthropic's Claude Mythos appears on Claude Code, finds 10,000 bugs
Technology
Anthropic's new AI, Claude Mythos, made a surprise appearance on Claude Code, just after the company said the model wouldn't be released to the general public anytime soon.
Despite being under wraps, Mythos has already helped spot over 10,000 major software bugs in just under a month and is accessible to select partner companies like Google and Amazon as part of Project Mythos.
Anthropic limits Mythos access to partners
Some users noticed models called "Mythos 1" and "claude-mythos-1-preview" pop up on Claude Code and Anthropic's security tool. These traces were quickly wiped away.
For now, Anthropic says only select partners (like governments and trusted companies) will get access until they're sure the tech is safe for everyone.