Anthropic's 'Mythos' AI sparks alarm

Mythos represents a significant leap in artificial intelligence, showing a 'step change' in its capacity to hack critical infrastructure.

Anthropic has stated the model is too powerful for general public release and is currently being used for stress-testing national security systems.

The development has energized the cybersecurity industry in China and the U.S., as professionals scramble to adapt to a new era of agentic AI. Despite being blacklisted from some government work earlier this year, Anthropic's latest breakthrough has made the company impossible for policymakers to ignore.