Key Takeaways:
Bill Gates, who once predicted AI would free humanity from work, now says the technology has crossed five danger thresholds with no plan to manage the fallout.
Bill Gates says AI has crossed five danger thresholds — from bioterrorism to job-market destruction — and society is unprepared, urging new taxes and protected human jobs in a 5,784-word essay published Wednesday.
"We've crossed the threshold in terms of AI's bio-capabilities, cyber-capabilities, psychosocial capabilities, job-market-destruction capabilities, and even the lack of control," Gates said in an interview with MIT Technology Review. "I'm just stunned at the lack of concern and discussion outside of the industry."
The essay, posted on his personal website, proposes two policy ideas: "human-reserved" jobs — occupations society deliberately keeps for people even where AI could perform them — and taxes on AI tokens and robots. Gates puts the ceiling for protected roles at 40 percent of the labor market, naming childcare and jury service as candidates. He expects customer support, software engineering, and paralegal work to vanish soonest, with construction and hospitality shifting toward automation by the end of the decade.
The warning carries weight because Gates has been among AI's most prominent optimists, predicting the technology would free people from work and accelerate medical breakthroughs. His reversal — "I'd love to be convinced that I'm wrong about the job market, but I'm not" — could reshape policy debates in Washington and Brussels, where AI regulation remains fragmented.
Gates's "human reserve" concept borrows from nature conservation: just as some land is protected from development because something there is worth preserving, certain jobs would be shielded from automation. He names childcare and jury service as roles humans are clearly better suited to, and suggests education and healthcare could receive selective protection rather than blanket exclusion.
The mechanism he proposes is tax rather than prohibition. Gates argues for taxing AI tokens and robots to reduce the financial incentive to automate, on the grounds that the current system "nudges you toward replacing people with machines" by taxing labor more heavily than capital. He is candid about the scale: "I'm talking about a change to the tax system that's greater than any in my lifetime."
A robot tax is not new — Gates floated one in 2017 to a cool reception from economists who argued it would tax productivity rather than protect workers. What has changed is the number of people who now expect to need it. Swiss research has found fewer job advertisements for career starters, and a Harvard study found AI-native startups hiring fewer juniors and more senior people.
Gates acknowledges the objections himself. Who qualifies for a protected job, who decides, how it is enforced, and what happens when a country that does none of this competes with one that does — all questions he raises without answering. He also notes that "you can't count on an industry to self-regulate," pointing to OpenAI, Google DeepMind, and Microsoft as companies whose leaders express private concern but continue racing ahead.
Gates's most chilling warning concerns biological capabilities. "Any model that can make novel molecules should be monitored," he said. "I view bioterrorism risk, versus a natural pandemic, as about 50 times more scary, more likely than a natural pandemic risk."
He proposes that the US require monitoring for any model capable of generating new molecules, and that Washington approach China on a joint agreement. "You don't even have to do it. You have to say what you're planning to do," he said. "I have no reason to think the Chinese won't go along."
Gates also said he was stunned by the pace of coding advances in the past year — citing Anthropic's Claude Code and agentic approaches — and realized months later that the same capability represented a massive cyberattack threshold. "And you know what happened as a result of that? Not much."
The essay is the first of multiple planned memos. Gates said a dedicated biosecurity memo will follow before the end of the year. He also acknowledged his own limitations as a messenger, citing his wealth, his history with Epstein, and his divorce. "I'm an imperfect messenger," he said. "Let's find the perfect messenger, and I'll share all my thoughts with that person."
For investors, the essay signals potential policy risk for AI companies. If Gates's proposals gain traction — particularly the token tax, which he suggests could capture 50 percent of revenue from AI usage that replaces human work — the economics of automation could shift materially. Microsoft, OpenAI, Google, and Anthropic all face potential regulatory exposure, though no concrete legislative action has been proposed in any jurisdiction.
This article is for informational purposes only and does not constitute investment advice.
