Anthropic's Mythos model raises alarms

Mythos is reportedly capable of shifting the balance between cyber attack and defense, prompting a private conference call between government officials and CEOs from Google, Microsoft, and OpenAI. The Bank of England is also set to join the discussion, examining how the tool could impact financial institutions.

While Anthropic has invited moral thinkers and religious leaders to help steer the model's behavior, experts fear the 'Vulnpocalypse'—a scenario where such advanced tools become widely available to hackers.

The company's momentum has made it a central focus of the HumanX AI conference in San Francisco.