OpenAI released a new report called "Industrial Policy for the Intelligence Age."

The headline pitch: companies should trial four-day work weeks. Governments should fund retraining. AI productivity gains should flow to workers. Everyone wins, apparently.

Pause on that for a second.

The company whose entire commercial strategy depends on selling enterprises software that does the work of junior staff is now telling those same enterprises to keep paying their workers for fewer hours. While also buying more OpenAI subscriptions. This is not industrial policy. This is a sales brochure with a think tank cover.

BBC covered the report on April 20. Grace Sheraton Neff at Oxford offered the polite version: "ideas might work but need complete political change." Translation: cute, not happening, next.

Look at the Sam Altman policy evolution. In 2024 he floated a robot tax. Specific, mildly costly to OpenAI, killed quietly. Then came universal basic income, which conveniently punted the entire labor question to governments. Now we have "industrial policy," which reframes the entire AI labor shock as an opportunity for civic leadership.

Each version of the pitch gets cheaper for OpenAI and more expensive for everyone else.

Notice what is not in the report. A commitment from OpenAI to fund any retraining. A pledge to revenue-share with displaced workers. A pricing carve-out for public institutions. A promise to delay any model release that displaces a verified percentage of a workforce. Nothing. Zero dollars of OpenAI money attached to OpenAI's four-day work week vision.

This is classic externalized cost. OpenAI gets the enterprise contract. Workers get the schedule change. Governments get the training bill. Taxpayers get the invoice.

There is also a convenient timing element. OpenAI is quietly pulling back on consumer products and pivoting harder into enterprise. Enterprise buyers care about two things: productivity gains and legal cover. A widely circulated policy report that reframes AI adoption as socially responsible labor reform is magnificent legal cover. When a Fortune 500 HR team tells the board they are cutting forty percent of a function and adopting OpenAI, the board now has a narrative. OpenAI said this was industrial policy.

The deeper play is reputation management ahead of the inevitable IPO. Microsoft spent years building its "responsible AI" posture before Azure became the dominant enterprise story. OpenAI is compressing that same arc into eighteen months. Expect more reports. Expect more keynote appearances next to policymakers. Expect Altman at Davos.

None of this is evil. It is just not industrial policy. Industrial policy requires the company driving the disruption to pay into the transition. Germany did this with its coal regions. Denmark did it with wind. Real industrial policy has capital commitments attached.

OpenAI wrote a paper.

The four-day work week is a great idea. Workers deserve more of the upside from automation. Retraining absolutely needs to be funded. All of it is true. None of it should be coming from the company selling the pickaxes.

If OpenAI actually believed its own report, page one would list the dollar figure OpenAI is contributing to the transition fund. It does not. So we will not either.

Sources: BBC (April 20, 2026), OpenAI "Industrial Policy for the Intelligence Age" report.