Bill Gates has sharpened his warning on artificial intelligence. In a long-form GatesNotes essay published on August 25, 2026, the Microsoft co-founder writes that the “turbulent AI era is here,” while governments still lack a credible plan for managing its economic, social, and political consequences.
This is not a conversion from AI booster to doomer. Gates continues to see the technology as a tool for improving medicine, education, agriculture, and scientific research. His argument is that AI development is moving faster than the institutions needed to distribute those benefits, protect children, manage job losses, and prevent dangerous uses.
That distinction matters. Gates is not calling for an end to AI development. He is warning that leaving the transition almost entirely to technology companies and market incentives could turn AI into what he describes as the “worst source of injustice.”
Gates Says the AI Transition Has Outrun Policy
Gates’s position has become more urgent, but the underlying optimism is familiar. In his 2023 essay, “The Age of AI Has Begun,” he compared the technology’s importance with the graphical user interface, the microprocessor, the internet, and the mobile phone. He also anticipated disruption in employment, education, privacy, and misinformation.
The new essay changes the emphasis. Gates now argues that societies have spent too much time debating whether AI will be transformative and too little time preparing for the transformation. He organizes the problem around five areas: employment, security, children, inequality, and international relations.
Some coverage has presented his scenarios as fixed predictions. MediaPost’s report on the essay focuses on his demand for government planning, while Billionaires Africa’s account emphasizes disappearing jobs, criminal misuse, and childhood harms. Gates’s original text is more careful: it presents possible trajectories and policy options, not a precise timetable or finished legislative program.
The essay also does not claim that technology executives are secretly terrified while publicly suppressing their concerns to protect investment. Gates criticizes inadequate planning and the concentration of power, but his target is institutional unpreparedness rather than an alleged conspiracy among AI leaders.
The Warning Spans Jobs, Security, and Childhood
Employment is the most immediate part of Gates’s case. He expects AI systems to take over a growing share of office work, beginning with routine and entry-level tasks. As robotics improves, he believes automation will also spread through construction, hospitality, manufacturing, and other physical industries.
AI adoption is already moving into ordinary workplaces. Gallup’s workplace tracking found that 26% of U.S. employees used AI at work at least a few times a week in the fourth quarter of 2025, up from 12% in 2024. That does not prove mass unemployment is imminent, but it shows how quickly the technology can become part of existing jobs.
Gates also points to national-security risks. Advanced systems may help attackers discover software vulnerabilities, conduct more sophisticated cyberattacks, or acquire knowledge relevant to biological and chemical weapons. OpenAI’s updated Preparedness Framework includes high and critical capability thresholds for biological and chemical threats, cybersecurity, and AI self-improvement. Those categories describe capabilities that require safeguards; they do not establish that today’s public models can autonomously execute every feared attack.
For children, the concern is less dramatic but already visible. Gates cites risks from deepfakes, automated cheating, persuasive misinformation, reduced independent thinking, and emotional attachment to AI companions. His policy prescription includes stronger age verification, limits on engagement-maximizing designs, and rules preventing companies from treating children’s attention and emotional dependence as products to monetize.
Human-Reserved Work Is Radical, but Not a 40% Quota
One of Gates’s most unusual proposals is to reserve parts of the economy for human labor. He identifies care professions as a likely candidate because trust, empathy, accountability, and physical presence are central to the work. He also argues that people should retain the option to interact with a human in areas such as healthcare and education.
The essay does not propose legally reserving 40% of all jobs for people. No such figure appears in Gates’s published plan. His actual position is narrower and less defined: governments should consider protecting selected sectors and services where replacing humans would damage social relationships or deny people meaningful human contact.
Even that narrower proposal would be difficult to implement. A hospital, for example, could use AI for scheduling, transcription, diagnostic support, and administrative work while still employing human doctors and nurses. Policymakers would need to determine which tasks require a person, rather than trying to label an entire occupation as either human or automated.
An AI Tax Would Change the Economics of Automation
Gates also wants governments to reconsider how they tax labor. Current systems collect substantial revenue from wages and payroll. If machines perform more economically valuable work while employing fewer people, public revenue could decline just as demand for unemployment support, retraining, healthcare, and income assistance increases.
His proposed answer is some form of tax on AI activity. That could mean taxing tokens, queries, automated outputs, or work performed by robots. The goal would be to shift part of the tax burden from human labor to machine-generated economic activity and use the proceeds for wage insurance, lifelong education, shorter working hours, or a stronger safety net.
This is a policy direction, not a ready-to-use tax code. Counting tokens would be especially difficult because models use different architectures, companies run systems internally, and one high-value output may require fewer tokens than a low-value task. A practical system might have to tax revenue, productivity gains, or automated capital rather than raw model usage.
The tradeoff is unavoidable. Taxing AI too heavily could slow valuable medical and scientific work. Taxing it too lightly could reward companies for replacing employees while shifting the social costs onto workers and governments. Gates’s contribution is to put that incentive problem at the center of the discussion.
Global Rules Require U.S.-China Cooperation
National laws cannot fully contain systems, model weights, biological knowledge, and cyber tools that cross borders. Gates therefore argues for international AI governance informed by civil aviation standards, nuclear inspections, and the Montreal Protocol, which coordinated global action against ozone-depleting chemicals.
Such a system could establish shared safety evaluations, incident-reporting requirements, restrictions on dangerous capabilities, and procedures for monitoring high-risk AI facilities. It would not require every country to adopt identical technology policy, but it would require agreement on minimum safeguards.
The hardest part is U.S.-China cooperation. Gates argues that neither superpower can independently control a technology being developed around the world. Reuters reported that he has discussed the issue with political leaders and met senior Chinese official Liu Haixing in Beijing in July 2026. Neither his essay nor Reuters’s report announces an upcoming Gates summit with Chinese President Xi Jinping.
Gates Still Wants AI Deployed
Gates’s alarm has limits. He believes governments should regulate high-risk systems while accelerating beneficial applications, particularly in countries that have historically received new technology late.
The Gates Foundation is already supporting AI projects in healthcare, agriculture, and education. Its work includes tools designed to assist health workers, give farmers localized advice, and translate educational material into more languages. Gates’s concern is not that AI lacks useful applications. It is that access to those applications may follow existing patterns of wealth and power unless governments and philanthropies intervene.
That makes his position more demanding than either unconditional enthusiasm or a blanket call to pause. He wants faster deployment where AI can reduce disease and poverty, combined with stronger restrictions where it can displace workers, manipulate children, enable surveillance, or increase access to destructive capabilities.
Final Thoughts
The most important part of Gates’s warning is not another prediction about how many jobs AI might eliminate. It is his recognition that technical progress does not automatically produce a fair social outcome.
His proposals remain incomplete, particularly the token tax and human-reserved sectors. Yet they move the discussion toward decisions governments can actually make: who pays for displacement, which services must remain human, what duties developers carry, and which risks require international inspection. The alarm is credible precisely because Gates still expects AI to deliver enormous benefits. His case is that those benefits will not distribute or govern themselves.
Frequently Asked Questions
4 questions
1What did Bill Gates say about AI in 2026?
Bill Gates said the world has entered a turbulent AI era without an adequate plan for managing job displacement, security threats, childhood harms, inequality, and geopolitical competition. He continues to support beneficial AI development, but argues that governments need stronger safety standards, labor protections, ethical limits, and international coordination.
