HELP WE TRAINED OUR AUTOCORRECT ON STORIES ABOUT AI THREATENING PEOPLE AND NOW OUR AUTOCORRECT IS GENERATING OUTPUT BASED ON ITS INPUT PLEASE THINK WE'VE MADE SOMETHING THAT IS CAPABLE OF THOUGHT PLEASE WE'VE GOT NOTHING ON DECK FOR WHEN THE FUNDING STOPS
Reposted from
Phil Lewis
An Amazon-backed AI model tried to blackmail engineers who threatened to shut it down, a safety report revealed
In tests, Anthropic's Claude Opus 4 would resort to "extremely harmful actions" to preserve its own existence
In tests, Anthropic's Claude Opus 4 would resort to "extremely harmful actions" to preserve its own existence
Comments