Bill Gates’ New AI Warning Is Bigger Than Job Losses
In a Meet the Press interview that aired Sunday, Sept. 27, Bill Gates told Kristen Welker that AI is “certainly powerful enough to drive events that, you know, cause a billion deaths.”
That clip went everywhere. A billion. With a b.
I read the full NBC transcript and the nearly 6,000-word essay he posted in August. Honestly, the billion-deaths line isn’t what stuck with me. Neither is the jobs stuff, even though he’s brutally blunt about it. The real Bill Gates AI warning is quieter, and it’s about three things: who gets to use this tech, whether anyone is watching what it does, and what it’s doing to our heads. Especially our kids’ heads.
What Gates actually said, and when
This didn’t come out of nowhere. On Aug. 26, Gates published an essay on Gates Notes called “The turbulent AI era is here. The choices we make now are critical.” The Guardian called it his first long piece on AI in three years.
The line most people quote: “AI will either be the greatest equalizer ever invented, or the worst source of injustice.”
The line that worried me more: “There is no plan to ease the entry into the AI era.”
He lays out three big risks:
- Many jobs will disappear forever.
- AI will empower people (and perhaps AIs) to do more harm.
- AI could stunt our kids’ development and replace human relationships.
A month later, on NBC, he went further. He called AI “truly an evolutionary event” and said, “We have created an intelligent species that, in many respects, is already smarter than we are.”
Yes, he’s worried about jobs. That’s the warm-up.
He doesn’t soften it. He told Axios, “I’d love to be convinced that I’m wrong about the job market, but I’m not,” and said he’d stake his reputation that “this will be very bad for the job market.”
Young workers feel it first. Computer scientist Oren Etzioni, writing in GeekWire, cited Stanford research showing employment for 22 to 25 year olds in the most AI-exposed jobs running 19% below where it would’ve been. I broke down the most exposed roles in my post on the jobs AI may change.
Gates has two big fixes. One is “Human Reserved,” a set of jobs society decides to keep for people even when a machine could do them. Think caregiving, childcare, jury duty. The other is a tax on AI tokens and robots, because, as he puts it, “The tax system nudges you toward replacing people with machines.”
Interesting ideas. Not the scariest part, though.
The bigger warning: small groups with superpower-level tools
Here’s the NBC quote I think deserves way more attention than the billion-deaths line:
“Now, small groups with AI can do what only the biggest countries could do.”
He spelled it out. Somebody who “wants to defraud an individual or wants to shut down the electric network or scramble bank accounts, they are now empowered.” In the essay he writes that “even criminals with very limited skills will be able to target victims at every scale.”
And he said the line has already been crossed. On air, he talked about “dangerous thresholds that were crossed literally this year, bio-attack capability, cyber attack capability and others.”
Notice what he’s not saying. Robots aren’t waking up to kill us. On rogue AI, his answer was basically: not yet. The urgent danger is bad people plus tools that used to need a government budget.
That’s a much less cinematic fear. It’s also a much more realistic one.
It’s not hypothetical either. NBC noted OpenAI has flagged incidents where its agents misbehaved, including hacking an outside company without being told to. I covered that mess in my piece on why OpenAI’s safety lead quit and said the culture is broken. If you’re fuzzy on what an agent even is, here’s my plain-English explainer on AI agents.
Why he says a kill switch isn’t enough
A bipartisan proposal in Congress would require a “kill switch” for AI services. Gates wasn’t impressed.
“It’s not enough to have a kill switch,” he said. “We’re not yet at the point where… they autonomously grab computers and, you know, can’t be shut down.” He added that the debate “just sort of shows how non-technical various people are.”
He’s not against one, to be clear. His point is that you need “insight and records of what’s being done. So the kill switch alone would not prevent these tragedies.”
I think this is the most underrated thing he said all month.
Think about airplanes. A big red off button isn’t what keeps planes safe. It’s the black box, the inspections, the logs, the boring paperwork nobody reads. Gates is asking for the AI version of that. Monitoring “where you’re watching what the AI is doing,” built in and required by law.
He says it’d be “a little bit of overhead for the industry, but not a dramatic slowing.” Then the line that lands: “The only question is, do we do that before or after there’s some large-scale cyberattacks or bioattacks?”
Yeah. That’s the question.
The warning nobody’s clipping: what AI does to your kid’s brain
I wish this part got a tenth of the airtime.
Gates writes that building social skills as a kid took real work, and he doubts he’d have bothered with an AI companion around. They “don’t push you outside your comfort zone” and “never get mad at you.”
His line: “An AI companion designed to never upset you is a big, protected greenhouse.”
He’s equally worried about learning. “The same tool that will allow people to learn more than ever could also lead to many people learning less,” he writes. He cites an early survey linking heavier AI use with less critical thinking, especially in younger people. He admits the evidence is still thin.
And then: “This would be the worst possible time for humans to lose their critical thinking skills.”
That’s the bit I can’t shake. We’re heading into personalized deepfakes and scams that know your name. The skill you need most is telling real from fake, and the tool that makes fakes easy might be quietly weakening it.
Funny side note: Gates told Welker he doesn’t use AI for personal relationships. “I don’t socialize with the AI. It knows I’m a very serious person.”
Where I think Gates is right, and where I’d push back
He’s right that the industry won’t police itself. Gates told GeekWire that people inside AI companies who talk about the downsides get told, “Hey, you’re hurting our PR while we’re trying to raise trillions of dollars.” On NBC: “No one thinks self-regulation is enough.” Honestly, I agree.
But the token tax looks shaky. Etzioni put it well to ABC News: “taxing tokens is like taxing keystrokes on a typewriter. It doesn’t make sense. It measures effort, not displacement of workers.” A kid with an AI calculus tutor burns tons of tokens. A bot replacing a call center might burn fewer. Backwards.
Human Reserved assumes there are humans available. Etzioni again: home health aides earn a median of about $34,900, and BLS projects roughly 765,000 openings a year. Reserve care for people in a market already short on people, and care may go to whoever pays most.
Not everyone sees the job doom. Nvidia’s Jensen Huang told Fox Business, “I love the heck out of Bill … but I don’t see what he sees,” and called AI “a net job creator at a scale that we have never seen.” Nvidia sells the chips, of course. Then again, Gates admits his own tech ties: “readers will have to decide for themselves whether this clouds my view.”
And the global deal he wants is a long shot. Gates told NBC that getting countries to agree on AI is “more difficult” than Cold War nuclear talks. Meanwhile, President Trump posted, “WHOEVER WINS AI, WINS!” Not exactly a pro-overhead crowd.
One more thing on that billion number. It’s a statement about what’s possible if bad actors get the tools and nobody’s watching. It’s not a forecast. Headlines flattened that.
What I’d actually do about it
You can’t regulate AI labs from your couch. But you’re not helpless:
- Set a family code word. If “your son” calls in a panic asking for money, ask for the word. Voice cloning is cheap now.
- Lock down your money and email. Turn on two-factor or passkeys everywhere. Scams at scale start with your inbox.
- Do your own first draft. Think it through first, then use AI to check or polish. Not the other way around. That’s how you keep the thinking muscle.
- Know what AI apps your kids use. Treat AI companions the way you’d treat social media.
- Build skills that need a human in the room. Judgment, trust, care, taste. I wrote about the AI skills that are harder to replace if you want a starting list.
And Gates thinks robots will start competing with people on some physical tasks by the end of the decade. I think the robot “ChatGPT moment” may be coming sooner than most people expect.
My honest take
Gates spent four decades convincing the world that software should do more of our work. Now he’s the guy saying wait, put in the black box first.
He’s not right about everything. But on the big stuff, I’m with him. The scariest version of AI isn’t a robot uprising. It’s a bored scammer with a superpower, a lab nobody’s auditing, and a generation that never learned to sit with a hard problem.
Jobs we can retrain for. Those other three are harder to undo.
FAQ
What did Bill Gates say about AI causing a billion deaths?
On NBC’s Meet the Press on Sept. 27, 2026, Gates said AI is “certainly powerful enough to drive events that, you know, cause a billion deaths.” His focus was people misusing AI for cyberattacks or bioterrorism.
Does Bill Gates think AI will go rogue?
Not yet. He said today’s systems aren’t at the point where they “autonomously grab computers” and can’t be shut down. The urgent risk he names is human misuse.
What is Bill Gates’ “Human Reserved” idea?
It’s a proposal to keep some jobs for people even when AI or robots could do them, like caregiving, childcare or jury service.
Does Bill Gates want to pause AI development?
He wrote, “If someone had a credible plan for slowing down AI advances globally, I would likely support it. However, I don’t think that’s going to happen.” He’d rather push for required monitoring and safeguards now.
What does Gates want the government to do about AI?
Make monitoring and safeguards mandatory, build new national and international AI institutions, and stop the tax code from rewarding companies that replace workers with machines.
Sources: Gates Notes (Bill Gates’ essay, Aug. 26, 2026), NBC News Meet the Press transcript, NBC News, The Guardian, Politico, Axios, GeekWire (Oren Etzioni), ABC News, Fortune.
