Artificial intelligence (AI) firm Anthropic says testing of its new system revealed it is sometimes willing to pursue "extremely harmful actions" such as attempting to blackmail engineers who say they will remove it.
Anthropic's new AI model turns to blackmail when engineers try to take it offline
-- TechCrunch (@techcrunch.com) May 22, 2025 at 1:56 PM
[image or embed]
Drudge Retort Headlines
Trump Fires Commissioner of Labor Hours After Weak Jobs Report (71 comments)
South Carolina Family Terrorized (44 comments)
Everybody Hates Trump Now (28 comments)
Smithsonian Removes Trump from Impeachment Exhibit (23 comments)
Jack Smith Under Investigation by Office of Special Counsel (22 comments)
Job Growth Slows Sharply Steep Revisions to Prior Months (21 comments)
Trump Reveals Plan to Revoke Birthright Citizenship (19 comments)
Trump Revives Presidential Fitness Test (18 comments)
'Color is neither a poison nor a crime' (17 comments)
Dozens Taken to Hospital After Flight Emergency Landing (15 comments)