The Most Dangerous AI Ever Built Was Cracked by a Discord Group on Day One
中文摘要
Anthropic的危险AI首日即遭破解
English Summary
Anthropic's restricted "Mythos" AI was cracked on day one by a Discord group who guessed its URL, despite being limited to partners.
Original Excerpt
Anthropic restricted Mythos to twelve handpicked partners. Someone guessed the URL before the press conference ended. Continue reading on Generative AI »