OpenAI Proposes Four-Day Week and Robot Taxes

On April 6, 2026, OpenAI released a policy brief, Industrial Policy for the Intelligence Age: Ideas to keep people first, proposing a suite of government and corporate actions to prepare for large‑scale economic disruption from advanced AI. The paper calls for time‑bound pilots of a 32‑hour/4‑day workweek at full pay, taxation reforms that move the tax base away from labor and toward capital/automation, a nationally managed public wealth fund seeded in part by AI companies, and taxes tied to automated labor.

OpenAI frames these measures against a near‑term technical trajectory in which AI systems materially increase productivity and, in some cases, outperform humans at complex tasks. The document treats productivity gains as convertible into shorter workweeks or direct citizen dividends rather than solely corporate profit. It also flags second‑order systemic risks—heightened cyberattack capability and biological misuse of models—that require broader governance and resource planning, including grid upgrades to support the electricity demands of more and larger data centers.

• •Four‑day workweek: The brief explicitly recommends incentivizing employers and unions to run time‑bound 32‑hour/4‑day pilots with no loss in pay while holding output and service levels constant. OpenAI proposes government incentives for employers to run these experiments. (OpenAI PDF; India Today)
• •Taxation and robot taxes: OpenAI suggests modernizing the tax system by shifting reliance away from payroll taxes and exploring taxes related to automated labor or machinery that displaces human work. The aim is to preserve public revenues as labor income shrinks. (AOL; India Today)
• •Public wealth fund: The proposal includes creating a publicly managed fund that captures AI‑driven growth and distributes returns to citizens—analogous to sovereign wealth funds. (Axios; AOL)
• •Infrastructure and energy: The brief recommends accelerating electricity‑grid expansion and coordination to absorb the rising energy needs of data centers used for training and running large models. (AOL)
• •Framing and urgency: Sam Altman positions the package as a starting point for national debate, warning of acute threats such as cyberattacks and biological risks as models become more capable—he told Axios, "I think that's totally possible" in reference to a major cyber threat within a near timeframe. The brief states, "We're beginning a transition toward superintelligence: AI systems capable of outperforming the smartest humans even when they are assisted by AI." (Axios; OpenAI PDF)

These proposals directly affect how organizations budget for headcount, benefits, taxation exposure, and infrastructure. For ML teams and product leaders, the push for shorter workweeks and productivity‑linked benefits reframes ROI calculations: measured productivity gains from model deployment could be diverted to employee time or benefits rather than scaling headcount reductions. For infra and MLOps engineers, the explicit call for grid and data‑center planning signals growing regulatory and public scrutiny of energy consumption, which will influence procurement, deployment cadence, and transparency requirements. For policy‑minded practitioners, the robot‑tax and public wealth fund ideas create potential new compliance and reporting burdens on automation metrics and model impact measurement.

• •Pilot outcomes: Which governments and large employers accept incentives to run 32‑hour pilots, and what metrics they publish on output, retention, and well‑being. (OpenAI PDF)
• •Legislative moves on taxation: Drafts that redefine taxable bases or introduce charges tied to automated output or model usage will directly affect corporate financial modeling. (AOL)
• •Public wealth fund mechanics: Whether proposals specify valuation methods for AI contributions, seeding mechanisms, or governance structures—these details determine feasibility and investor response. (Axios)
• •Infrastructure policy: Regulatory or investment announcements accelerating grid upgrades or placing constraints on data‑center siting and energy sourcing.

OpenAI positions this package not as final policy but as an agenda setter to spur debate and early experiments. For ML practitioners, the document signals that technical choices—what and how you automate, how you measure productivity, and your energy footprint—are likely to become policy levers and compliance signals in the next 12–36 months.