The AI Researcher Who’s 10% Sure He’ll End Humanity—and Cashes His Paycheck Anyway

One of the most influential people in artificial intelligence right now believes there’s a better than 1-in-10 chance his life’s work will kill every single person on Earth. He said so, out loud, to a BBC reporter. And then he went back to his desk at Anthropic to keep building it.

Let me make sure you feel this. Because the quote from that researcher isn’t a confession. It’s a business card.

“I earnestly believe we’re on track to end all human life in a decade, but you better believe I’m gonna keep collecting this Anthropic check.”

Read that again. There it is: the entire tragedy of our species, compressed into twenty words and a direct deposit.

You’ve probably worked a job you hated. But have you ever worked a job you believed would incinerate your family, your neighbors, and every unread email in your inbox? Have you ever looked at a codebase and thought, this could be the last thing I ever write—for anyone—and then sat through a morning standup discussing sprint velocity?

Because that’s the reality of AI in 2025. The people building the future have already calculated the odds, done a quick mental cost-benefit analysis, and decided that a 10% chance of human extinction is a reasonable price for a career in tech.

But here’s the twist that should genuinely terrify you.

The only thing more terrifying than an AI researcher who thinks the world ends in 2035 is one who has already made peace with it.

This isn’t naivety. It isn’t ignorance. It’s rational. It’s structurally encouraged. And once you see it, you can’t unsee it.

Because think about the word “warning.” When an AI researcher says “this technology could kill all humans,” what does that sound like to an investor? It sounds like power. It sounds like a product so disruptive that it rewrites the rules of reality. When the CEO of a company you’re funding casually mentions that the model you’re training might bring about the eclipse of the human race, that’s not a safety alert—that’s a pitch. “We’ve built something so potent that it could end the story of our species.” Do you know how many trillions of dollars are being raised on that exact sentence?

Doom isn’t a warning. It’s a brand.

And the researcher? He gets to feel like a tragic hero in a Greek drama—the man who carries the fire even as it burns him. He gets a high-status job, a million-dollar salary, and the comfortable conviction that he’s “telling the truth.” Meanwhile, the truth isn’t changing anything. It’s part of the hype cycle.

Let’s be honest about the incentives. If you truly believed that your daily work would extinguish every life on the planet within a decade, you would quit. You would leak documents. You would expose the company. You would chain yourself to the server racks and scream until someone unplugged them.

But nobody does that. Why? Because the person who walks away loses everything—the money, the status, the sense of being at the center of history. The person who stays gets a paycheque and a front-row seat to the apocalypse. And if the apocalypse never comes? They’re a hero. If it does? They were “brave enough to warn us.” It’s a win-win.

Except for us.

The comment sections under the BBC article tell you everything. “Cool to see the BBC writing about this,” says one. Another casually drops: “Gentleman, the Great Filter.” Look at how calm everyone is. Look at how normalized it’s become to say “yes, we might all die, but anyway…”

The only acceptable response is to stop treating AI researchers as prophets and start treating them as what they are: well-paid workers in a dangerous industry.

When a coal miner says the mine might collapse, you don’t give him a raise. You evacuate. When a pharmaceutical exec says the drug might kill people, you don’t invest more. You recall it.

But when an AI researcher says “there’s a 10% chance I’m building humanity’s final chapter,” we nod, write an insightful LinkedIn post about existential risk, and then ask what the next funding round looks like.

There is a movement trying to change this. Groups like PauseAI are organizing, marching, demanding that we actually treat AI extinction risk as a red line instead of a sexy talking point. They’re dismissed as Luddites, as sci-fi nerds, as people who don’t understand “the opportunities.”

But let me ask you a question. Who is acting more sane?

The researcher who says “I think there’s a better than 10% chance this kills everyone, but my stock options are good, so I’ll keep going”?

Or the people who look at that 10% and say, “Wait, what?”

If a man told you he had to spin a roulette wheel where one chamber in ten would incinerate your children, you wouldn’t call him a visionary. You’d call the police.

We’ve surrendered our future to a group of people who have convinced us that the most rational thing they can do is keep building a machine they may not be able to stop—all while telling us, with a straight face, that they know exactly how dangerous it is.

That’s not a safety warning. That’s a threat.

And the worst part? More than 10% of you reading this will do absolutely nothing about it. Because you, like the researcher, have already made your peace. You’ve decided that the odds are good enough. That someone else will handle it. That being alive in 2035 is probably fine.

Probably.

But “probably” is not a plan. And “I’m cashing the check” is not a moral stance. It’s an epitaph.