The Internet Is Not Okay After This Anthropic Employee Shared That He Quit His Job Over Fears That AI May Kill Us All

Posted 1 hour ago

On Tuesday night, 27-year-old AI researcher Jacob Coxon posted a series of tweets that went megaviral, explaining his decision to quit his job at Anthropic, known for the AI model Claude.

Jacob is not the first in the industry to warn about the potential consequences of AI. Earlier this week, OpenAI chief scientist Jakub Pachocki published a blog post in which he wrote, "I am concerned no one is prepared for the consequences of a continued rapid rise in machine intelligence."

But Jacob's warning is particularly disturbing. In his viral thread, he writes, "The people building AI earnestly believe that it could kill us all by the end of the decade." As someone who would like to live to see 2031, that doesn't sound great!

In response, a lot of people are like, "Hey, please don't kill us all, thanks so much!"

At this point, you might be wondering, if so many people who work in AI are concerned about its ramifications, why do they continue doing what they do? According to Jacob, it comes down to competition.

Or, as this Twitter user very eloquently summed it up, "If I don’t build the Dick Smashing Machine, someone else will build the Dick AND BALL Smashing Machine…"

Jacob ended his thread with a callout to other AI researchers to think carefully about what they are doing. He writes, "Should you put your head down because 'it's happening anyway' - or take this moment to call for different conditions?"

Evan Hubinger, alignment science lead at Anthropic, also chimed in on the thread, claiming that "we really do earnestly believe AI could kill all humans!" Evan went on to clarify that he thinks the risk of current AI models wiping out humanity is "low," but he worries about how future models capable of self-improvement might behave.

In case you didn't know, AI alignment is a field of study that looks into how we can teach AI to behave as ethically, helpfully, and safely as possible. 

Another Anthropic employee, scalable oversight lead Samuel Marks, also chimed in to confirm that AI developers are concerned about the tech leading to "human extinction." In his tweet, Samuel shared that he signed an open letter alongside many other AI industry workers, urging the industry to slow down and asking for the US government to "support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development."

President Donald Trump, who loves a late-night AI slop posting spree more than perhaps anyone else currently alive, has advocated for slashing regulations around AI.

But in response to Jacob's posts, other lawmakers have signaled their willingness to take on regulating AI, including Senator Bernie Sanders.

But others are more skeptical about Jacob's claims. Some X users have hypothesized that he's hyping up the potential harms of AI as a marketing tactic. And others, including tech journalist Taylor Lorenz, seem to think it amounts to nothing more than "sanctimonious doomer posting."

And then, of course, there are plenty of people making darkly funny jokes about whatever all of this is, because that's the only way to cope. Here are a few of my favorites:

Sings "Unwritten" menacingly:

A relatable dilemma:

A reassuring text:

Back in my day, we used solar-powered calculators under our desks in math class, and it was fine!

An important question:

Don't worry, everyone, Anthropic's (not real, very satirical) Director of Killing People, Adolph Bundy, says that everything is going according to plan.

We've reached out to Anthropic and OpenAI for comment, and will update you if we hear back. If you need me in the meantime, I will be crawling around the BuzzFeed office unplugging all the computers.

Comments