Alongside concerns about safety, the race to advance artificial intelligence has triggered fears that the technology will wipe out millions of jobs — particularly white-collar ones — and concentrate wealth in the hands of those with a financial stake in AI models, chips and data centers.

Policymakers, economists and tech leaders are debating how to prevent these outcomes — or cushion the fallout if it comes to that. Researchers say artificial intelligence is already reducing hiring of entry-level roles in certain fields. And the AI boom has created enormous wealth for those invested in it.

The pressure to head off or find solutions that could offset the potential effects of AI-driven job losses and inequality may be especially strong in the US. The country hosts a high concentration of the most valuable companies developing AI and supplying related chips. And though the threat of job displacement applies to workers all over the world, unemployment benefits are less generous in the US than in many other wealthy countries. Here’s a primer on proposed measures.

Under this framework, citizens directly get a portion of the wealth generated by AI companies. Advocates say that because the technology was trained on data contributed by the public, people deserve to share in gains created.

Some proponents of a dividends program envision funds coming from a redistribution of AI companies’ profits or revenue through taxes. Kim Yong-beom, who was then the policy chief of South Korea’s president, said in a social media post in May that the nation could pay citizens a “dividend” using excess tax revenue generated by the AI boom.

But right now, leading AI companies are generating little or no profits, and may not do so for years, even as the technology is adopted widely around the world. Governments may not be prepared to impose big taxes on a loss-making industry if that continues, even if many jobs are eliminated.

At the same time, the companies’ valuations are enormous. An alternative to redistributing a portion of profits or revenue to the public via taxes would be to give people an ownership stake in AI companies. Anthropic PBC and OpenAI have advocated for versions of this idea. US President Donald Trump has said Americans can benefit from the success of AI by the government taking stakes in tech companies. “If we do that, the public will become very rich,” he said in June.

Vermont Senator Bernie Sanders has proposed giving the public a share of AI’s riches through a sovereign wealth fund. Under a bill he introduced in June, large AI companies would be required, on a one-time basis, to relinquish 50% of their equity to such a fund. It would pay an annual dividend of 5% of its value to support direct payments to Americans and other measures to ensure a high standard of living. This plan would allow regular citizens to benefit directly from the companies’ rising value, Sanders writes. “If the big AI companies continue to grow as rapidly as many analysts expect, then the value of the sovereign wealth fund will grow as well.”

While government officials debate mandatory redistributions, some AI entrepreneurs are proposing voluntary giveaways. In mid-October, Anthropic announced a plan to put 5% of its shares into a fund to promote the “public good,” for example, through donations to nonprofit organizations or direct payments to people. Addressing “labor market impacts” was named as a possible area of focus. The new fund, which is subject to stockholder approval, would be a wholly owned LLC — rather than a nonprofit — and would grow as the company grows, according to Anthropic.

It would be created as a public benefit fund, a new type of entity designed to go beyond achieving financial returns to make a positive social or environmental impact. The company added that its co-founders had earlier pledged “to donate” 80% of their stock holdings, without saying to what or by when.

Hours after Anthropic made its announcement, Elon Musk pledged to give one of his shares of SpaceX to every child in the US. While SpaceX is mainly in the aerospace business, it is the parent company of xAI, Musk’s artificial intelligence venture.

Another commonly discussed intervention if AI reduces the number of jobs is a basic income program — cash payments given directly to citizens from government coffers.

Historically, many cash-payment programs have targeted specific groups, such as one in Kenya designed for rural people and another in Finland for the unemployed. The Stanford Basic Income Lab has documented more than 200 pilot projects in the US alone. Funded through a mix of philanthropy and public funds, these programs typically provide modest income for a fixed period of time to dozens or hundreds of members of groups such as single mothers or former prisoners. Some are designed to cover basic needs broadly; others help pay for childcare or job training.

In the US, a group called Fund for Guaranteed Income in March launched one of the first such programs for workers displaced by AI. The nationwide project, which targets workers in roles including tech and marketing, provides about $1,000 in cash per month over a year as well as skills training and job coaching to a small group of participants.

In August, President Lai Ching-te of Taiwan — which has a world-leading semiconductor industry — announced plans to make one-time cash distributions to eligible residents in 2027, citing the country’s surging economic growth fueled by global demand for AI. The payments are small — NT$10,000 ($314) per recipient. Taiwan also distributed cash to residents in 2025.

Last year, the Marshall Islands became the first country in the world with a nationwide universal basic income program. The program uses a blockchain-based currency for payment. Other countries have toyed with providing universal basic income: South Korean President Lee Jae Myung advocated for such a program when he ran for office, citing potential job displacement from AI and other reasons.

Universal basic income programs in the US have been proposed over the years. Musk touted the idea about a decade ago in response to rising automation. The proposal was popularized by tech entrepreneur Andrew Yang during his 2020 Democratic presidential campaign.

Champions of income programs say they help cut poverty and point to government emergency payments distributed during the Covid-19 pandemic as proof that cash transfers are effective. Studies published in 2024 by OpenResearch, a project backed by OpenAI co-founder Sam Altman, found that recipients spent more on meeting their basic needs and assisting others than before.

Even some libertarians, including Charles Murray, a scholar at the think tank American Enterprise Institute, say cash-transfer payments allow recipients rather than the government to choose what to do with the money and can replace the patchwork of welfare programs.

But basic income programs can be expensive to fund, and critics say they don’t address the sense of purpose and identity that many people derive from work. Conservative critics say these programs are a waste of taxpayer funds and discourage people from working. Finland’s cash-transfer program drew scorn from the country’s biggest labor union for being counterproductive because it encouraged people to work less.

After years of promoting the idea of a universal basic income, Musk has since 2023 promoted “universal high income,” which would allow people to secure “whatever goods and services they want,” as he put it in a post on his social media platform. In other words, guaranteed income, but high instead of basic.

The billionaire entrepreneur hasn’t shared many details about how that would come about. In a social media post in April, he said that AI will cause economic output to grow faster than the money supply, resulting in deflation, and that universal high income paid “via checks issued by the Federal government is the best way to deal with unemployment caused by AI.”A co-founder of an Israeli insurance company unveiled a model earlier this year that he describes as the “plumbing” behind Musk’s talk of high-income payments.

The model assumes AI will keep growing the economy even as it replaces jobs, allowing governments to redirect part of the resulting windfall profits and extra tax revenues into a guaranteed income for everyone.

Altman for years championed universal basic income and funded large cash-transfer experiments in the US. But he recently said he’s less convinced about the value of such programs. Altman is now advocating for what he calls “universal basic compute,” where instead of money, people would receive a portion of AI computational power. (“Compute” is a term used in AI and cloud computing to mean the resources devoted to computation.)“Everybody gets a slice of GPT 7’s compute and they can use it, they can resell it, they can donate it to somebody to use for cancer research,” he said.

“What you get is not dollars but this slice; you own part of the productivity.”He’s vague on details. Altman has not explained how such a system would work in practice, including whether people — or possibly public institutions such as schools — would receive transferable credits that represent an amount of computing power, or whether they would receive some other form of claim on AI-generated productivity.

In June, former US Secretary of Commerce Gina Raimondo and former Indiana Governor Eric Holcomb co-founded a group, RAISE US, that’s focused on helping US workers adapt as the use of AI expands in the economy. RAISE US seeks to raise $1 billion to create pilot versions of programs to retrain and redeploy workers in industries such as healthcare and advanced manufacturing or help them start their own businesses. Raimondo said that while the US spends billions of dollars on higher education, workforce training and unemployment insurance, little is tied to whether those programs produce good results. The Labor Department in April moved toward embedding AI skills into existing apprenticeship programs.

Whether retraining programs are effective is the subject of debate. Though a partner of RAISE US, Anthropic said in an August paper that large government job training programs haven’t historically delivered lasting gains on employment and earnings.

Another possibility is wage insurance programs, which would provide income to workers who are displaced by AI and have to take jobs with lower salaries.

David Autor at the Massachusetts Institute of Technology is among the labor economists who have championed the idea. Under such a program, the government would temporarily cover the difference between a worker’s old salary and the lower pay in their new job. Autor argues that wage insurance should complement, rather than replace, retraining, and might last up to a few years for a worker.

Proponents of the idea have largely been vague about how the determination would be made that AI had forced a worker to take a job with inferior pay.

This would not be the first wage insurance program in the US. A federal program called Trade Adjustment Assistance was created in the 1970s for workers who had lost their jobs after their employers shifted roles overseas. The program was phased out in 2022. Another idea: using changes to the tax code to incentivize companies to invest in human capital.

The Mercatus Center, a research group at George Mason University, calls for reforming aspects of the federal tax code that it says incentivize investments in technology over human labor. For example, the tax code allows businesses to claim a tax deduction for giving an employee additional training for their current role, but not for a new one. The Mercatus Center proposes allowing tax deductions for training that prepares workers for new roles, too, which would encourage businesses to train their workers for roles that demand AI.

“In the pre‑AI era, this distinction was benign,” the groups says, but today, denying deductions for new-role training “penalizes firms that teach frontline staff how to collaborate with large language models,” the group says.

One of the world’s largest hedge funds, Bridgewater Associates, said it backs the introduction of a “token tax” that would be levied on companies’ AI consumption as a way to help level the playing field between work done by humans and the technology. Human labor is currently at a disadvantage to AI, because employers pay payroll taxes for workers, while work carried out by computers isn’t taxed at all, the fund said. AI is generally treated as a capital expenditure that can qualify for tax deductions and depreciation write-offs. Bridgewater said a 35% token tax could raise as much as $600 billion by 2030.

In January, Democrats introduced a bill in New York’s state legislature that would create a surcharge to be levied on some companies that lay off a specified number of employees as a result of AI. The fee would be at a rate of 2% of the company’s income and would be paid to the state’s labor department to be used for retraining programs or an unemployment insurance fund.