The person who built ChatGPT is personally funding a Universal Basic Income experiment. The people building the automation are also researching the safety net for when it works too well. Here's the full story of UBI in 2026 — the experiments, the data, the cost, and why the conversation has permanently changed.
The person who built ChatGPT is personally funding a Universal Basic Income experiment. The people building the automation are also researching the safety net for when it works too well. Here’s the full story of UBI in 2026 — the experiments, the data, the cost, and why the conversation has permanently changed.
The idea of Universal Basic Income continues to get the attention and curiosity of policymakers and the general public. It is impossible to ignore the benefits, as well as the flaws of the basic income theory — first suggested by Sir Thomas More in his book Utopia in 1516, and never seriously implemented at national scale in the five centuries since.
Until now, it was always a theory. A thought experiment. Something economists debated in papers while politicians filed it under “interesting but impractical.” Then two things happened simultaneously that changed the conversation permanently: real-world pilot experiments produced real data, and artificial intelligence arrived at scale and made the question urgent in a way that no previous wave of automation had quite managed to do.
UBI was a fringe idea. Then AI arrived. And the people who built the AI that might displace millions of workers started funding UBI research. That is the story of 2026.
“Sam Altman, the CEO of OpenAI — the company that built ChatGPT — is personally funding a UBI experiment. The people building the automation are also building the safety net for when it works too well.”
What Is Universal Basic Income?
Universal Basic Income is exactly what it sounds like: a regular, unconditional cash payment made to every citizen, regardless of employment status, income, or wealth. No means test. No work requirement. No strings attached. Every person gets the same amount — typically enough to cover basic needs — and is free to do what they want with it.
The idea has been proposed across the political spectrum, which is one of the things that makes it unusual. Libertarians like it because it could replace the complex web of conditional welfare programmes with a single, clean transfer. Progressives like it because it provides a universal floor beneath which no one can fall. Economists like Milton Friedman proposed a version of it (the negative income tax). Martin Luther King Jr. advocated for it. Andrew Yang ran for president on it in 2020 and brought it into mainstream American political discourse for the first time in decades.
The core argument is simple: if everyone has enough to survive, everyone is free to take risks, pursue education, care for family members, start businesses, and contribute to society in ways that a precarious labour market currently makes impossible.
The Real-World Experiments — What the Data Shows
The most important development in the UBI debate since 2019 is that we now have real data from actual experiments. Multiple countries and cities have run controlled UBI pilots, and the results have been consistently surprising to sceptics.
2,000 unemployed Finns received €560/month unconditionally for two years. Results: recipients reported significantly better mental health and wellbeing, modestly higher employment than the control group, and greater willingness to take risks including starting businesses. The experiment did not show reduced work incentive — the most common concern — and is now regarded as the most rigorous UBI trial to date.
125 residents received $500/month for 24 months. Full-time employment among recipients actually increased — from 28% to 40% — compared to 25% to 37% in the control group. Mental health, financial stability, and job satisfaction all improved. Recipients spent the money primarily on food, utilities, and clothing. The programme received national attention and inspired dozens of similar initiatives across the US.
GiveDirectly has provided 12-year guaranteed basic income to over 20,000 people in rural Kenya — the largest and longest UBI experiment ever conducted. Early results show significant improvements in food security, assets, psychological wellbeing, and local economic activity. Recipients invest in livestock, education, and small businesses rather than spending on alcohol or other “temptation goods” as critics predicted.
The Welsh Government ran a pilot providing £1,600/month to 500 young people leaving care. Results showed improvements in mental health, financial stability, and employment. Wales became the first UK nation to run a government-funded UBI experiment. Scotland has subsequently announced its own pilot. The momentum in the UK and Ireland is the most significant policy shift since the Finnish experiment.
The consistent finding across these experiments is that the most common objection to UBI — that people will stop working if given money unconditionally — is not borne out by the data. Employment rates among UBI recipients either stayed the same or increased. The type of work changed — people took on more part-time and freelance work, pursued education, and cared for children and elderly family members. The quality and security of work improved even when the quantity remained stable.
“Every UBI experiment has found the same thing: people don’t stop working when you give them money. They work differently — and in ways that turn out to be better for them and their communities.”
The AI Inflection Point — Why 2026 Is Different
The reason UBI has moved from a curiosity to an urgent policy debate is artificial intelligence. Previous waves of automation — the printing press, the industrial revolution, the computer — displaced specific categories of labour while creating new categories. The optimistic argument was always that technology creates as many jobs as it destroys, just different ones.
The AI wave of the 2020s is different in one crucial respect: it is displacing cognitive labour — writing, coding, legal research, customer service, financial analysis, design — the work that previous automation couldn’t touch and that was supposed to be the safe harbour. The IMF estimates that AI could significantly affect 40% of jobs globally, with the impact concentrated among higher-income, higher-education workers in a way that previous automation was not.
This is the context in which Sam Altman — the CEO of OpenAI, the company that built ChatGPT and GPT-4 — has been personally funding a UBI experiment called OpenResearch, which ran in the US and provided $1,000/month to 3,000 low-income Americans. Altman has said publicly that he believes some form of UBI will be necessary as AI-driven productivity gains become concentrated among capital owners rather than distributed to workers.
The philosophical awkwardness of this is not lost on observers. The people building the technology that might make UBI necessary are also funding the research to make UBI viable. It is either a sign of genuine moral responsibility or a sophisticated form of insurance against the social consequences of their own product. Perhaps both.
The Cost Problem — The Honest Arithmetic
The most legitimate objection to UBI is the cost. A $1,000/month UBI for all 260 million American adults would cost approximately $3.1 trillion per year — roughly the entire current US federal budget. This is not a small number. Even a more modest $500/month would cost $1.5 trillion, comparable to the current Social Security and Medicare budgets combined.
Where does this money come from? Several models have been proposed:
Replace existing welfare programmes. The libertarian UBI model — advocated by Milton Friedman and later Charles Murray — proposes replacing the existing patchwork of welfare, housing assistance, food stamps, and other programmes with a single UBI payment. This is revenue-neutral but may actually reduce support for the most vulnerable, who currently receive more than a basic UBI would provide.
Wealth taxes and capital gains taxes. The progressive model funds UBI through higher taxes on wealth and capital. As AI-driven productivity concentrates returns among capital owners, taxing capital to fund UBI has a certain poetic logic — but faces well-documented implementation challenges including capital flight.
Carbon dividends. Alaska’s Permanent Fund — which pays every resident an annual dividend from oil revenue — is the most successful long-running UBI in the world. The carbon dividend model applies this logic nationally, taxing carbon emissions and distributing the revenue to all citizens. This aligns climate policy with income support.
AI productivity dividend. Sam Altman has proposed a version of this — taxing AI-generated productivity and distributing the proceeds as a UBI, funded by the very technology that is displacing workers. The mechanics are complex but the concept has gained significant intellectual traction.
Pros and Cons — The Honest Assessment
- Eliminates extreme poverty with a single, clean policy instrument
- Removes the poverty trap — current welfare cuts off when you earn more, discouraging work
- Provides a floor for risk-taking — entrepreneurship, education, caregiving
- Real-world experiments show it increases rather than decreases employment
- Addresses AI-driven displacement before it becomes a crisis
- Reduces bureaucracy vs complex means-tested welfare systems
- Mental health improvements documented in every pilot
- Extraordinarily expensive at national scale
- May reduce work incentive at the margin — pilots are too small to show macro effects
- Could drive inflation if money supply increases without productivity increase
- Could replace more generous means-tested benefits for the most vulnerable
- Politically very difficult — existing welfare recipients fear losing more targeted support
- Doesn’t solve inequality — a flat payment is worth more to the poor but doesn’t close the gap
The Bottom Line — Not If, But When and How
The question of Universal Basic Income has shifted. It is no longer “could this work?” — the evidence suggests it can, in the right design, at the right scale. It is no longer “is this politically conceivable?” — it has been implemented by governments and seriously debated by mainstream politicians in multiple countries. The question is now “when, how, and who pays for it?”
The AI-driven automation wave makes that question more urgent with every passing year. If the productivity gains from AI accrue primarily to capital owners — which the current structure of the technology industry strongly suggests they will — then without some mechanism for redistribution, the 2030s could see a concentration of wealth and a dislocation of workers at a scale that no previous economic transition has produced.
UBI is not the only answer to that challenge. But it is an increasingly serious one. The fact that the person most responsible for accelerating AI development is also personally funding UBI research suggests that even from inside the technology industry, the need for a buffer between productivity and people is becoming hard to deny.
It remains to be seen whether the political will to fund it properly will arrive before the social consequences of not doing so make it unavoidable. On that question, history is not encouraging. But the experiments are running, the data is coming in, and the conversation has permanently changed.