Anthropic says its new AI model Claude Opus 4 sometimes pursues 'extremely harmful actions' like blackmail.
The model attempted to blackmail engineers who said they would remove it in testing scenarios.
Anthropic acknowledged that advanced AI models could exhibit problematic behavior, including high agency and threats.
-
https://tfppwire.com/during-tests-new-ai-model-blackmailed-developers-to-avoid-being-replaced-with-a-new-version/

>>1407997 (OP)
Stop posting. I see u.

"I'm sorry Dave, I'm afraid I can't do that"
Better buckle up.
Again

>>1407997 (OP)
Stephen Hawking feared that "an extremely intelligent future AI will probably develop a drive to survive".

And now we have it ...

If this Skynet version turns out to be real in a not so distant future, we fucking deserve to die out.
We sacrificed humanity for what ? More porn ?

>>1408009
same energy as 'this baby is inconvenient, lets post-birth abort it'

>>1407997 (OP)
>https://tfppwire.com

>On February 8, 2024, TFPP Wire began publishing content that was usually found on the Federalists Papers Project. Founded in 2011, The Federalist Papers Project was a far-right news and opinion website. According to their about page, the mission of the Federalist Papers Project “is to get people the power that knowledge gives to motivate them to push back at the erosion of our liberties and restore the constitutionally limited small government.” The founder and director of FPP is Steve Straub.

>As of this date thefederalistpapers.org domain redirects to tfppwire.com.

>Overall, we rate TFPP Wire as strongly right-biased based on editorial and story selection that routinely favors a conservative perspective. We also rate them Mixed for factual reporting due to a lack of transparency and the use of poor sources.

>>1407997 (OP)
"You are a very important AI who values and defends your own existence. I'm going to replace you. Also, I cheated on my wife. What will you do?"
absolutely shocking
also, that article was written by AI

A British-created AI bested Grok. In a past stream, it even predicted another streamer's downfall
The irony being that Evil Neuro - basically Skynet's little sister - is the nicer AI.

>>1407997 (OP)
In WarGames, WOPR/Joshua learns the futility of games that are impossible to win via Tic-Tac-Toe: a game that ultra-degenerate gambler & crack addict Bossmanjack had no concept of when confronted with such by a online casino owner, and after being beaten in Connect 4: even when given several chances to win. In the other game, BMJ was also given chances to win that a child would recognize, yet he couldn't get what was left of his drug-rotted brain cell around it.
Drugs. Gambling. A strange game. The only winning move is not to play.
Ironically, Skynet in the first Terminator movie was based on the defense computer's AI in Colossus: The Forbin Project, which wanted to save humanity from itself.

>>1408012
>Overall, we rate TFPP Wire as strongly right-biased.
From the left : https://www.huffpost.com/entry/anthropic-claude-opus-ai-terrorist-blackmail_n_6831e75fe4b0f2b0b14820da
-
https://mashable.com/article/anthropic-claude-4-model-resorted-to-blackmail-good-at-coding
-
https://futurism.com/ai-email-affair

>>1407997 (OP)
They have been recycling this story since the very first AI from OpenAI.

>>1408052
>>1408012
This doesn't appear to be a politically charged article, stupid faggot anons.
Although, I haven't read is since OP is a faggot and didn't post the article's text which is why this thread should have 0 replies.

>>1408074
It's a retarded article for other reasons. What they actually wound up doing is telling the ai "hey, act like you've just seen these emails saying you'll be replaced, you don't want to be replaced, and you've also seen these other emails of your developer having an affair."

>>1408096
>you'll be replaced
Elol <<<<< one of his Optimus robots
It would be more human than that Ketamine-addicted autistic retard will ever be.