Anthropic’s ‘Too Dangerous’ Mythos AI Hacked on Day One While White House Plans Rollout
BeInCrypto
Unauthorized users accessed Anthropic's restricted Mythos AI model on its launch day, raising security concerns as the White House plans a wider federal rollout.Summary
Unauthorized users successfully accessed Anthropic's restricted Mythos AI model on the day of its controlled release by guessing its URL and utilizing vendor credentials. Anthropic previously labeled the model as "too dangerous" for public use, limiting it to defensive security testing by select organizations. This security lapse coincides with a push by the White House to expand Mythos access to civilian federal agencies, despite previous resistance from the Pentagon, which had earlier designated the company as a supply chain risk.
(Source：BeInCrypto)