Anthropic says newest AI model is too powerful to release to public

https://www.youtube.com/watch?v=AfdDYu43Cy0

Anthropic says its newest model, Claude Mythos Preview, is too powerful to release publicly because it can find and exploit serious software vulnerabilities with unusual accuracy. Instead of a full launch, Anthropic is giving limited access to a small group of partners for defensive cybersecurity work.

Why it matters

The company says the model’s capabilities raise concerns that criminals or spies could misuse it, especially for breaking into major operating systems and web browsers. Anthropic says it wants to eventually release “Mythos-class” models more broadly, but only after stronger safeguards are in place.

What’s different

This is one of the clearest public examples yet of a major AI company deciding a model is not safe for general release even though it is technically ready. The move underscores how quickly frontier AI is advancing and how hard it is for companies to balance capability with security.