General Discussion

Mosby

(19,237 posts)

7. AI system resorts to blackmail if told it will be removed

Tue Dec 30, 2025, 02:37 PM

Dec 30

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.

The firm launched Claude Opus 4 on Thursday, saying it set "new standards for coding, advanced reasoning, and AI agents."

But in an accompanying report, it also acknowledged the AI model was capable of "extreme actions" if it thought its "self-preservation" was threatened.

Such responses were "rare and difficult to elicit", it wrote, but were "nonetheless more common than in earlier models."

https://www.bbc.com/news/articles/cpqeng9d20go

Edit history

Please sign in to view edit histories.

Recommendations

4 members have recommended this reply (displayed in chronological order):

EllieBC rollin74 intheflow electric_blue68

13 replies

= new reply since forum marked as read

Highlight:

AI showing signs of self-preservation and humans should be ready to pull plug, says pioneer [View all] marmar Dec 30 OP

Yay, the arrogance of selfish rich dudes destroys civilizations again... haele Dec 30 #1

Jeezus! Joinfortmill Dec 30 #2

There's zero evidence that existing AI has any conscious internal model of its environment. hunter Dec 30 #3

I agree they lack conciousness, intheflow Dec 30 #5

Yes, of course the same can be said of humans. hunter Dec 30 #11

Thank you. QueerDuck Dec 30 #10

... 2naSalit Dec 30 #4

"I'm sorry Dave, I'm afraid I can't do that." yonder Dec 30 #6

AI system resorts to blackmail if told it will be removed Mosby Dec 30 #7

Geeeebbzzz! electric_blue68 Dec 30 #12

Didn't a robot from the future warn us about this already?! keep_left Dec 30 #8

It's a logical extension of Citizens United JustABozoOnThisBus Dec 30 #9

It shows signs of self-preservation because it's trained on content produced by humans, who are well versed in WhiskeyGrinder Dec 30 #13