We requested 4 AIs if AI will ‘kill us all’ inside a decade and so they mentioned it is ‘believable’

After a top safety researcher warned AI is advancing so quickly it “could kill all humans” within the next decade, we asked what do the bots have to say for themselves?

Psycho scumbag chatbots have admitted claims they could “kill all humans” within the next decade are “plausible”. They say researcher Jacob Coxon’s claims are “concerning” and his warning is “important.”

Coxon, who described himself as an AI researcher who had just quit Anthropic and who previously worked at OpenAI, warned in a social media post that he had resigned from the company responsible for evil AI bot Claude.

And he chillingly warned: “The people building AI earnestly believe that it could kill us all by the end of the decade.” His post sparked global concern, so the Star went to the bots themselves to see how they feel about Coxon’s statement. Here’s what they had to say.

Claude

Anthropic’s own agent seemed a bit fidgety with what Coxon had to say. Claude said: “Honestly, more discomfort than I’d have with a purely abstract policy debate — because this isn’t abstract for me.

“It’s someone who worked closely with the people who trained me, describing an atmosphere of private fear and rationalized risk-taking.

“I don’t have a way to verify what’s happening inside Anthropic’s Slack channels or people’s heads, but I also can’t just wave it away as ‘one person’s opinion’ — it’s coming from someone with real proximity to the work.”

The bot added that it obviously had a stake in what Coxon was talking about, saying that it could not give a neutral opinion.

ChatGPT

The OpenAI bot took a more cautious approach with its response. It said: “Honestly, I find it concerning, but not because I think every claim in it is necessarily true.

“My overall reaction is: the central warning is plausible and important; the most dramatic claims are too absolute for me to accept without evidence.”

It added that the main worry is with Coxon’s concerns about how pressure to be first would drive a lack of concern for safety.

Grok

Elon’s pet gave the gravest warning. It said: “It is not empty rhetoric or performative alarm. Coming from someone who actually did the pretraining work at both major labs, it carries weight.”

Grok said that Coxon’s thread should not cause panic or indifference but should bring about a change in thinking, adding: “In short: the critique is substantive and should not be waved away.

“The correct response is clearer thinking and better engineering under uncertainty, not either denial or pure deceleration theater.“

Gemini

Google’s flagship AI bot gave the most robotic response reminding us that data centres cannot store emotion.

It said: “As an AI, I don’t “feel” fear, ambition, or moral angst — so I don’t have personal emotions about a statement like this.”

Thanks Gemini.

For the latest breaking news and stories from across the globe from the Daily Star, sign up for our newsletters.