Bill Gates says the people building artificial intelligence know exactly how dangerous it is and are staying quiet because saying so out loud would threaten the funding it relies on.
In a nearly 6,000-word essay published on his GatesNotes website, the Microsoft co-founder did something he has never done in five decades of championing new technology: He said he wishes it would advance more slowly.
Gates told The New York Times that tech executives privately discuss AI’s dangers but suppress those concerns because raising alarms publicly would jeopardize the capital pipeline fueling development.
He described the silence as deliberate, financially motivated, and industry-wide. The essay names three risks Gates says warrant immediate action.
The silence Gates says the industry is keeping
The sharpest section of the essay targets the people building the technology, not the politicians trying to regulate it. Gates described the pattern as structural, a function of how AI capital moves.
Founders who voice safety concerns risk being read as ambivalent about their product, and ambivalence, in a market this competitive for funding, is expensive. Gates himself is insulated from that pressure in a way current AI CEOs are not.
Three AI risks Gates says need action now
- The first risk is permanent job displacement, with entry-level and mid-level workers likely to face the earliest losses. Customer service, software, legal support, and sales could be hit first, followed by blue-collar jobs as robots become cheaper. He singled out medicine and law as fields where AI is now capable of displacing credentialed professionals.
- The second risk is widespread safety threats, as AI makes dangerous capabilities easier to access. Small groups with limited expertise could launch cyberattacks, conduct fraud, or pursue biological weapons that once required state resources.
- The third risk is psychological harm to children, particularly as AI companions become increasingly engaging. Always-agreeable AI companions could foster addictive behavior among young users and displace meaningful real-world friendships.
Stanford offers early evidence showing AI’s impact on young workers
Gates grounded his employment warning in a Stanford Digital Economy Lab study analyzing millions of payroll records from ADP, the nation’s largest payroll processor.
Employment among workers aged 22 to 25 fell roughly 11% in the two most AI-exposed occupation groups. Meanwhile, employment for the same age group rose about 10% across the three least-exposed occupation groups.
More AI:
- Nvidia just made a move Wall Street wasn’t ready for
- Microsoft just took sides in the AI policy fight
- OpenAI just disclosed something genuinely alarming
Together, those trends leave employment among 22- to 25-year-olds in highly AI-exposed occupations roughly 19% below expected levels, up from 15% a year earlier.
The Stanford researchers noted that they do not see widespread, economy-wide job displacement from AI.
Declines are concentrated in fields where AI directly substitutes for human tasks, including software development and customer service.
Gates proposes taxing AI and reserving certain jobs for humans
Gates paired his warnings with three policy proposals that go beyond the voluntary commitments most AI companies have offered.
The first calls for new national bodies and international oversight modeled on nuclear inspections, aviation rules, and ozone treaties.
The second is a category he calls “Human Reserved,” a set of jobs society would protect from AI replacement even when the technology becomes capable of doing so.
He compared the concept to nature reserves, where communities could develop the land but choose not to because the loss outweighs any economic gain.
Gates told Axios that child care and jury service clearly fit that protected category, and he estimated roughly 40% of work could be reserved initially. The idea has roots in personal experience, as his father received around-the-clock care before dying of Alzheimer’s in 2020.
His third proposal is a tax on AI tokens and robots to level the cost gap between human and machine labor.
Employers pay payroll taxes on human workers, but purchasing a robot often qualifies as a deductible business expense, creating an incentive to replace people with machines, Gates noted.
An AI expert pushes back on Gates’ tax proposal
Not everyone agrees with Gates’ prescriptions, even among researchers who share his concern about AI’s trajectory.
Oren Etzioni, professor of computer science at the University of Washington, told ABC News that taxing AI tokens amounts to taxing keystrokes on a typewriter.
I think that taxing tokens is like taxing keystrokes on a typewriter. It doesn’t make sense. It measures effort, not displacement of workers
He also raised a competitive concern: If the United States taxes tokens while other countries do not, companies could shift usage to foreign-built models, effectively offshoring their AI activity.
Still, Etzioni stressed it is not too late to act on the risks Gates identified. Policymakers still have a narrow window to respond, he told ABC News.
What Gates’ AI warning means for younger workers and unprepared policymakers
If Stanford’s findings hold, young workers may already be entering a labor market shaped by a 19% employment gap. That shift could arrive well before any of Gates’ three proposals have a chance to reach a vote.
Congress has no active bill proposing an AI token tax or framework to protect jobs from AI displacement. There is also no international oversight body resembling the system Gates proposed for nuclear weapons inspections.
What happens next depends on three things: which industries cross the 19% threshold, whether the payroll-tax-versus-capex gap draws bipartisan interest, and whether any member of Congress is willing to say publicly what Gates says tech executives will only say privately.
Related: Bill Gates makes $180 million bet on backbone of America’s economy
