When jobs are lost, paychecks follow. Disappearing paychecks mean disappearing payroll taxes that fund Social Security, Medicaid and SNAP.
OpenAI warned that a chain reaction is already underway. The company released a 13-page policy paper calling on governments to tax automated labor and artificial intelligence (AI)-driven capital returns. Without intervention, OpenAI argued that payroll tax revenue will decline as AI replaces workers. That erosion puts Social Security, Medicaid and SNAP at risk. Taxes on automation, OpenAI said, can help close that gap without slowing AI adoption.
Taxing Automated Labor
AI expands corporate profits and shrinks labor income. Programs built on payroll taxes don’t adjust automatically. The Hill reported that OpenAI called for taxes “related to automated labor” to stabilize that funding base.
OpenAI’s paper also proposes higher levies on capital gains and corporate income. It pairs those with wage-linked incentives for firms that retain and retrain workers, modeled on existing R&D credits. However, the ChatGPT maker stopped short of specifying rates or a collection mechanism.
The proposals aren’t new to the industry. Anthropic’s economic policy research, published last October, identified similar instruments. University of Virginia economists Lee Lockwood and Anton Korinek proposed studying taxes on token generation, robots and digital services. They argued that taxing compute and hardware could become the only remaining mechanism to capture windfall gains if labor’s role in the economy declines.
Anthropic noted those taxes would directly affect its own revenue. The company has committed $10 million to scale up its Economic Futures research program, which funds empirical work on AI’s labor market effects.
Concentration and the Distribution Problem
OpenAI’s paper acknowledged that productivity gains will concentrate inside a small number of firms, including OpenAI itself. Workers using AI may see output rise. Their compensation may not follow.
The paper called for a public wealth fund seeded by AI companies and government investment. Returns would go directly to citizens, including those with no exposure to financial markets. It also proposed converting efficiency gains into better retirement contributions and expanded healthcare cost-sharing.
It called for time-bound pilots of a 32-hour workweek at no pay cut, with reclaimed hours converting to a permanent shorter week or bankable paid time off. AI systems have already progressed from tasks that take minutes to tasks that take hours, the company said. Systems capable of handling month-long projects are next, it added.
Credibility Questions
Tech Policy Press called the document a “policymercial,” a portmanteau of “policy” and “commercial.” It noted several sections read as product pitches rather than policy. The proposal asking policymakers to help workers launch AI-first businesses, for example, named no role for state or federal governments and cited no funding mechanism.
The paper also linked subsidized AI access for underserved communities directly to a call for governments to accelerate energy infrastructure expansion and reduce regulation on advanced conductors. OpenAI is among the largest consumers of that infrastructure.
TechCrunch reported that the proposal mixes traditionally left-leaning mechanisms like public wealth funds with a fundamentally market-driven economic framework. OpenAI was founded as a nonprofit premised on AI benefiting all of humanity. It became a for-profit company last year, a shift that’s led critics to question whether its stated mission is compatible with its fiduciary duty to shareholders.
Some analysts told Fortune the document correctly identifies AI as a structural economic shift rather than a technology problem. Carnegie Endowment visiting scholar Anton Leicht called the proposals fundamental societal changes and heavy political lifts.
Anthropic’s research puts the stakes directly: governments need to formulate responses before the scale of displacement is clear, because waiting means the tools won’t exist when they’re needed.
OpenAI is offering research grants of up to $100,000 and up to $1 million in API credits for work that builds on its proposals. It opens a policy workshop in Washington in May.
