Amazon-Backed AI Model Would Try To Blackmail Engineers Who Threatened To Take It Offline
In tests, Anthropic's Claude Opus 4 would resort to "extremely harmful actions" to preserve its own existence, a safety report revealed.
The company behind an Amazon-backed AI model revealed a number of concerning findings from its testing process, including that the AI would blackmail engineers who threatened to shut it down.
‘I Even Believe He Is Destroying the American Presidency’
25 May 25
One thing stands out amid all the chaos, corruption and disorder: the wanton destructiveness of the Trump presidency.
The targets of President Trump’s assaults include the law, higher education, medical research, ethical standards, America’s foreign alliances, free speech, the civil service, religion, the media and much more.
About 89% of the public want their governments to do more to tackle the climate crisis – but don’t know they’re the majority
The Guardian is joining forces with dozens of newsrooms around the world to launch the 89% Project – and highlight the fact that the vast majority of the world’s population wants climate action. Read more
