OpenAI published a policy paper this week titled "Industrial Policy for the Intelligence Age: Ideas to Keep People First." It proposes a robot tax, a public wealth fund, expanded social safety nets, and a government-subsidised four-day work week. The company filed this document from a position of having just closed a $122 billion funding round that may value it at close to a trillion dollars. The juxtaposition is worth sitting with for a moment.
The substance of the proposals is not trivial. The robot tax concept, revived from Bill Gates's 2017 suggestion, would require companies employing AI systems to pay tax equivalent to what they would have paid employing humans in the same roles. The public wealth fund would give Americans an automatic stake in AI infrastructure as a form of collective ownership of the gains from AI productivity. The four-day work week proposal would subsidise reduced hours at full pay as AI absorbs more routine labour. Taken together, the framework describes how to distribute AI-driven economic gains more broadly rather than concentrating them at the top of the capital structure.
TechCrunch described the document as "blending traditionally left-leaning mechanisms with a fundamentally capitalist, market-driven economic framework," which is a reasonable characterisation. OpenAI is not proposing to constrain its own development or price its products differently. It is proposing that governments build safety nets and wealth mechanisms around the disruption that OpenAI is accelerating. The company will continue to do what it does; the question of who benefits is passed to public institutions to resolve.
The internal contradiction is real but not necessarily cynical. OpenAI's leadership appears to genuinely believe both that AI displacement is coming on a large scale and that the right response is redistribution rather than restraint. This is the dominant techno-optimist position: the technology is inevitable, the benefits are real, the task is to share them. Bloomberg noted that the proposals also include calls for grid investment and expanded safety-net spending, suggesting the company sees itself as a stakeholder in the infrastructure required to operate AI at scale, not just a beneficiary of it.
There are two serious critiques worth making. The first is the obvious timing question: a company that just raised $122 billion proposing progressive economic policies is in a complex position when it comes to the question of whether those policies should apply to it. The robot tax, if implemented, would apply to OpenAI's customers using AI to replace workers, not necessarily to OpenAI itself as a software provider. The policy burden falls downstream.
The second critique is structural. The proposals assume the displacement is a given and focus on what to do in the aftermath. That framing skips the prior question: whether the particular kind of AI deployment driving mass job replacement is actually necessary, or whether it is a product of a market structure that could be organised differently. A company with genuine power to shape AI deployment arguing for downstream redistribution is making a choice about where in the system to intervene. OpenAI is not proposing to change the price structure that makes AI labour cheaper than human labour; it is proposing to tax the gains from that price differential after the fact. Whether that is the right level to intervene at is a question the paper does not engage with directly.
The four-day work week proposal connects to a long history of technology promising to free workers from drudgery. John Maynard Keynes predicted fifteen-hour work weeks by the year 2000, on the basis of then-current productivity trends. The hours never arrived: productivity gains were absorbed by consumption growth and profit extraction, not leisure. OpenAI's version of the same promise is genuinely well-intentioned, but the mechanism for actually delivering it, rather than just announcing it, is not spelled out.