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
'This Isn't a Christian Nation' (105 comments)
US Only Country Facing Tourism Decline (54 comments)
Don't call it 'Alligator Alcatraz,' Call it a Concentration Camp (31 comments)
GOP Rep Torched after Hospital in his District Shuts Down (28 comments)
RFK's Proposal to Let Bird Flu Spread Through Poultry Could Set Us Up for a Pandemic, Experts Warn (21 comments)
Trump Admin has Begun Garnishing Wages of Student Loan Borrowers in Default (19 comments)
UK: Ozzy Osbourne Sings Final Concert with Black Sabbath (18 comments)
US Trade Deficit Widens in May on Weak Exports (18 comments)
Social Security Administration Sends Misleading Email (18 comments)
Trump says he Plans to Host UFC Fight at the White House (16 comments)