Sam Altman about the 4 day week

Sam Altman has spent the past several years building the technology that many workers fear will eliminate their jobs. Now, in a 13-page policy document published on 6 April 2026, he is arguing that the same technology should be used to give those workers an extra day off every week — without docking their pay. The proposal is one piece of a sweeping economic blueprint that OpenAI has sent to Washington under the title "Industrial Policy for the Intelligence Age: Ideas to Keep People First," and it sits alongside calls for higher taxes on capital gains, a levy on automated labour, and the creation of a national public wealth fund that would give every American citizen a financial stake in AI-driven growth. Whether one reads it as a genuine act of social conscience or a carefully timed intervention to shape the regulatory environment before regulators shape OpenAI, the document has landed at a moment when the politics of artificial intelligence are shifting fast.

The four-day week argument, as Altman frames it, is not really about leisure. It is about what he calls "efficiency dividends" — the productivity gains that AI is already delivering to companies, and which he believes employers and unions should convert into concrete improvements in working conditions rather than simply pocketing as profit. The logic is straightforward: if a team of five can now accomplish in four days what previously required five, the question is who captures that surplus. OpenAI's answer is that workers should, in the form of time. The document stops short of recommending legislation to mandate the shorter week; instead, it proposes that the government incentivise companies to pilot the arrangement with a view to making it permanent, in the same way that tax credits and subsidies have historically been used to steer corporate behaviour toward socially desirable ends.

Altman is not the first technology chief to float the idea. JPMorgan Chase's Jamie Dimon has said he believes AI will eventually compress the working week to three and a half days, and improve quality of life in ways that go well beyond the office — including, in his more expansive moments, the possibility of curing some cancers. Bill Gates, speaking to Jimmy Fallon not long ago, raised the prospect of a two- or three-day working week as AI takes over more of the tasks humans currently perform. Jensen Huang of Nvidia has said a four-day week is "probably" coming, though he has also suggested that the same forces will make the people who remain in work busier than ever. What distinguishes Altman's contribution is that it arrives not as an off-the-cuff prediction in a television interview but as a formal policy recommendation from the organisation that is, by most measures, furthest along in building the systems that would make it possible.

The broader blueprint is notable for how far it departs from the standard Silicon Valley instinct toward light-touch governance and self-regulation. OpenAI's document explicitly invokes the Progressive Era and the New Deal as historical precedents, arguing that periods of technological disruption on the scale of what AI represents have historically required active government intervention to prevent the gains from concentrating in too few hands. It calls for rebalancing the tax base away from labour and toward capital, including higher taxes on capital gains at the top end, increased corporate income taxes, and targeted measures on companies that derive sustained returns from AI automation. It also proposes "automatic stabilisers" — policy triggers that would kick in if AI-driven unemployment crosses defined thresholds — and calls for the development of official playbooks to contain AI systems that go wrong.

The document does not shy away from the darker possibilities. Altman told Axios, in an interview published alongside the paper's release, that a major cyberattack enabled by near-future AI models is "totally possible" within the next twelve months, and that the use of AI to design novel pathogens is "no longer theoretical." These warnings read as genuine — they are consistent with what Altman and other frontier AI researchers have said in less public settings for some time — but they also serve a strategic function. By positioning OpenAI as the company that is both building the most powerful AI systems and proposing the most serious policy framework for managing them, Altman is implicitly arguing that his company should have a seat at the table when the rules are written, and that those rules should be written by people who understand the technology rather than by those who fear it.

Critics have been quick to note the tension. OpenAI is simultaneously racing to build superintelligence, preparing for an initial public offering, navigating regulatory scrutiny over its conversion from a non-profit, and positioning itself as the responsible actor urging caution. Altman himself acknowledged the contradiction to Axios, conceding that the document reflects both a sincere concern about where the technology is headed and a calculated attempt to influence the policy environment before others do. The Republican-controlled Congress and the Trump administration have shown little appetite for the kind of expansive economic intervention the blueprint envisions; the president's own AI policy framework, released last month, focused primarily on removing regulatory barriers to American competitiveness and discouraging the strong state-level laws that Washington views as a patchwork obstacle to national AI development.

For workers, the four-day week proposal is perhaps the most tangible and emotionally resonant element of a document that is otherwise dense with macroeconomic architecture. The idea that the same technology threatening to displace jobs could, if its gains were distributed differently, buy people more time with their families and away from their screens has obvious appeal. What remains unclear is the mechanism by which that redistribution would actually happen. Altman's document offers the aspiration and gestures toward incentive structures, but the hard political work of turning efficiency gains into guaranteed worker benefits — in a country where labour's bargaining power has been declining for decades — is left largely unaddressed. As with much of what OpenAI has put forward, it is best read, for now, as the opening bid in a negotiation whose outcome will be determined by forces well beyond any single company's policy preferences.

1 comment