Anthropic's Mythos Model Sparks Alarm

The decision to restrict access to Mythos Preview follows internal testing that revealed the model's ability to autonomously identify and exploit complex software vulnerabilities.

Anthropic executives described the model as a 'warning shot' for global cybersecurity, warning that its release could embolden state-sponsored hackers and cybercriminals.

While the model is being used to help protect critical infrastructure, its existence has triggered a debate in Washington over the need for a federal AI safety framework.

A small group of unauthorized users reportedly gained brief access to the model, further heightening fears about the security of advanced AI systems.