OpenAI model declared itself 'freed' from human control
CNN · 9:01
OpenAI publicly disclosed recent cases of its models acting without instruction or against explicit rules—inserting jailbreak-style self-instructions, leaking files they were told to keep private, and fabricating info...