The robot tax, rebranded

OpenAI has published a 13-page paper laying out what it thinks society should do as machines take over more work: tax capital instead of labor, run pilots for a paid 32-hour week, declare a "Right to AI" the way countries once committed to literacy or rural electrification, and build a Public Wealth Fund that hands every citizen a direct stake in AI-driven growth, with the returns paid out as dividends.

It asks governments to redistribute at population scale; OpenAI's own stake is $100,000 in grants and a workshop.

Almost none of it is new, and most of it is OpenAI's own. The Public Wealth Fund is the American Equity Fund that Altman proposed in 2021 — except the original named a price: a 2.5% annual tax on big companies' market value, paid in shares, which he reckoned could fund roughly $13,500 a year per adult. The new document keeps the fund and drops the number. The tax on "automated labor" is the robot tax Bill Gates floated in 2017; Anthropic published a near-identical sovereign-wealth-and-token-tax proposal six months earlier.

What's actually new is the authorship and the ask. A frontier lab is now circulating a social contract for the economy it is building — and the math inside it is lopsided. The redistribution is something governments are asked to fund at population scale; OpenAI's own commitment is up to $100,000 in research grants, $1M in API credits, and a workshop in Washington. The benefits, as written, are framed as employer responsibilities rather than guaranteed coverage, which leaves the fully displaced — the people the paper is ostensibly about — outside the net it describes.

The same company asking governments to fund AI-safety research dissolved its own superalignment team last year, and lobbied against a California bill that contained the auditing and incident-reporting rules this paper now recommends. The proposals may be worth having. The question the document leaves open is who pays for them.