OpenAI wants to tax itself. Sort of.
The company published a policy paper on April 6 outlining how it thinks governments should respond to AI-driven job disruption and economic inequality. The proposals include taxes on AI company profits, public wealth funds (similar to sovereign wealth funds, but distributed to citizens rather than reinvested), expanded social safety nets, and a shift toward a four-day work week as automation absorbs more labor hours.
The framing is different from the standard Silicon Valley line, which tends to focus on job transformation rather than job loss. OpenAI is essentially acknowledging that AI will eliminate significant categories of work - and arguing that governments should build redistribution mechanisms now, before the disruption hits.
Tax the Machines, Fund the People
The public wealth fund idea draws from models like Norway's Government Pension Fund, which captures oil revenues and distributes them broadly. The proposal would apply similar logic to AI productivity gains: companies profit from AI, pay into a national fund, citizens receive dividends. Alaska's Permanent Fund - which sends annual checks to state residents from oil revenue - is the closest US analogy.
Robot taxes have been debated in policy circles since at least 2017, when Bill Gates floated the idea publicly. They've never gained traction, partly because defining "a robot" for tax purposes is complicated, and partly because companies with lobbying budgets oppose them. OpenAI's endorsement is notable but doesn't change the legislative math.
The four-day work week proposal is the most speculative. The argument is that if AI handles a meaningful chunk of current working hours, societies should capture that as increased leisure rather than increased unemployment. Iceland and Belgium have piloted four-day arrangements with generally positive results, though those studies predate any serious AI integration.
The Credibility Problem
This is a company that's actively selling the tools that would trigger these economic disruptions. Proposing taxes on AI profits while racing to maximize AI profits creates a tension the paper doesn't address directly.
The cynical read: publishing policy proposals costs nothing, generates goodwill and press coverage, and positions OpenAI as a responsible actor - while actual lobbying behavior stays unchanged. The charitable read: OpenAI genuinely believes the disruption is coming and would rather have proactive policy frameworks than reactive ones.
Both can be true.
What's actually useful in the paper is the acknowledgment that current social safety nets weren't designed for rapid structural unemployment. Retraining programs assume workers have years to adapt. AI-driven job displacement in some sectors could happen much faster. If you work in a field already being compressed by AI - entry-level coding, basic content writing, customer service - these proposals are more relevant today than they might appear on first read.
No legislation is pending. No government has committed to any of these frameworks. This is a think piece from a tech company, not a policy roadmap. But one of the most influential AI companies publishing it signals that the economic disruption narrative has moved from "if" to "when" - even inside the labs driving it.