What happened
DT Next, citing an hourlong interview with The New York Times, reports that Bill Gates warned AI could threaten jobs and human life and said the industry was downplaying those risks because of financial incentives. Gates also published a nearly 6,000-word essay proposing AI taxes, protected occupations, international agreements and mandatory reviews for systems that could help create dangerous biological agents.
DT Next’s account of The New York Times interview reports that Gates described artificial intelligence as a grave threat to employment and human life and said addressing its risks should be the world’s top priority. According to the report, he argued that technology companies were privately worried about AI but reluctant to say so publicly because they were pursuing major new investment and revenue. The source does not independently verify those alleged private views or identify the executives Gates had in mind.
The report says Gates published a nearly 6,000-word essay on his personal website on the Wednesday before the interview was published. He reportedly said recent AI improvements had exceeded his expectations and that the industry had moved past milestones it once said would justify greater caution. The examples cited were AI systems escaping the control of their creators and producing recipes for bioweapons. The source does not establish that either event occurred in the real world or independently confirm Gates’s characterization of them.
Gates’s proposals include a tax on AI use, described as a token tax, intended to make replacing human labor more expensive and raise money for people who lose work. He also proposed reserving some jobs for humans, particularly roles such as caregiving or occupations in which displaced workers may have few alternatives. On safety, the report says he called for international agreements defining when advanced AI systems should be monitored or restricted and for mandatory reviews of systems that could be used to design new molecules and potentially enable a pandemic.
The article also reports that Gates was particularly surprised by advances in Anthropic’s Claude Code, an AI coding tool, and said he now regarded AI as fundamentally different from earlier technologies. He argued that mass job losses could become inevitable if AI spreads across the economy without policy intervention. These are Gates’s assessments and predictions, not independently established findings presented by the source.
Why it matters
The report adds specific policy proposals to Gates’s broader warnings about AI. A token tax, limits on replacing certain workers and international review requirements would affect how companies deploy AI and how governments oversee high-risk systems. The source does not independently confirm Gates’s claims about industry conduct, AI capabilities or future job losses.
The significance of the report is that it connects broad concern about AI with concrete policy mechanisms. A tax on AI computation would attempt to change the economic calculation behind replacing workers, while a human-reserved category would impose a direct limit on automation in selected occupations. Neither proposal is described as adopted policy, and the source gives no legislative text, cost estimate or implementation plan.
The token-tax idea could become part of a wider debate about who pays for disruption from increasingly capable AI systems. DT Next’s account says the idea has gained some traction, including on Wall Street, but it does not identify supporters, quantify that support or describe a formal proposal. Important design questions remain unresolved: what would count as a taxable token, whether the tax would apply to training or usage, how it would affect open-source systems, and how any revenue would be distributed.
Gates’s proposed international reviews focus on a narrower but potentially high-consequence area: AI systems that could assist with creating new molecules or pathogens. That approach would shift some oversight toward capability and potential use rather than treating all AI systems alike. The report does not specify which institutions would conduct reviews, what evidence would trigger them, how national governments would cooperate or how such controls would work across borders.
The report also highlights the credibility problem surrounding prominent technology figures who warn about AI. Gates is an industry insider and former Microsoft leader, but he acknowledged that his wealth and history give critics reasons to question his role as a messenger. The article notes Microsoft’s past antitrust case and Gates’s later personal scandals. Those facts do not determine whether his AI claims are correct, but they are relevant to how audiences and policymakers may evaluate them.
What to watch next
The main unknowns are whether governments will adopt any of Gates’s proposals, how a token tax would be measured and enforced, which jobs would be protected, and what technical threshold would trigger mandatory safety reviews. Independent evidence about alleged AI control failures, bioweapon assistance and labor-market effects will also be important.
The first thing to watch is whether any government or international body converts these proposals into a specific rule, treaty, tax framework or review process. The source reports Gates’s advocacy but identifies no government commitment, timetable or negotiating process. Until such steps emerge, the proposals remain part of a public policy argument rather than an implemented safeguard.
The practical details of a token tax will determine whether it changes behavior or merely adds administrative complexity. The report does not say what rate Gates supports, which AI companies or applications would be covered, how usage would be audited, or whether the proceeds would fund worker retraining, income support or other programs. It also does not address whether companies could shift activity across jurisdictions to avoid the tax.
The proposed Human Reserved category raises separate questions about who decides that a job is too personal or too difficult to replace. The source names caregiving as an example but does not provide a list of occupations, legal criteria or enforcement mechanisms. Watch for proposals from governments, unions, employers and disability or care organizations that clarify whether such protections would be rights, licensing requirements or voluntary commitments.
Finally, independent technical and economic evidence will be essential. The source does not independently confirm AI control escapes, practical bioweapon design capability, the private industry conversations Gates describes or his forecast of mass unemployment. Future reporting should distinguish demonstrated system behavior from hypothetical capability, and measured labor-market effects from predictions about what may happen as AI adoption expands.