Anthropic's Claude Mythos AI breaks weakened AES, outside cryptographers confirm
Technology
Anthropic's Claude Mythos AI managed to crack a weakened version of AES, a key system that helps keep digital data safe.
It took the AI just a week to figure out the attack, and outside cryptographers confirmed the results.
While today's full-strength AES is still secure, this shows how powerful AI could spot future flaws in even our most trusted encryption.
HAWK creators confirm Claude Mythos break
Claude Mythos also broke HAWK, a system designed to stand up against both quantum and regular computer attacks, with the creators themselves confirming it.
Before going public, Anthropic shared these findings with the US government and tech partners.
With AI moving so fast, experts say it's time to rethink and update our security standards to stay ahead.