Anthropic's Mythos AI security risks

The White House and industry leaders are racing to patch critical systems after Mythos demonstrated an unprecedented speed in identifying exploits that human hackers might take months to find.

While Anthropic has limited the model's public release, select firms are currently testing it to identify their own weaknesses before malicious actors can.

Canadian Artificial Intelligence Minister Evan Solomon is scheduled to meet with Anthropic leadership to discuss the need for trusted international security protocols.

The situation has prompted comparisons to the development of the atomic bomb, with critics arguing the world is now fundamentally less safe.