Six Counter-Proposals for the Intelligence Age
OpenAI released a 13-page industrial policy document and invited public feedback. We took them up on it. Six counter-proposals on workweek protections, healthcare, training data, public compute, tax modernization, and AI-enabled direct democracy.
OpenAI released a 13-page document called Industrial Policy for the Intelligence Age and invited public feedback. We took them up on it.
The document has real proposals worth engaging with: a Public Wealth Fund that would give citizens a direct stake in AI-driven growth, portable benefits that follow workers across jobs, and automatic safety net triggers tied to displacement metrics. Credit where it's earned.
But a 13-page industrial policy document that never mentions healthcare reform, never addresses training data compensation, and never uses the word "antitrust" has some gaps. Our response paper offers six counter-proposals that take OpenAI's own logic seriously and push it further than its authors were willing to go.
The six counter-proposals:
1. Work Week Protections With Teeth. OpenAI suggests 32-hour workweek "pilots." We propose a federal 32-hour standard with overtime protections, four weeks guaranteed vacation, and 10 paid holidays. The US is the only OECD country that mandates zero paid leave. Iceland and the UK have already run the trials. The evidence is in.
2. Healthcare Decoupled from Employment. You can't write 13 pages about restructuring the American economy and leave healthcare as a bullet point about "portable benefits." The employer-healthcare link is a World War II accident, not a design choice. If AI disrupts employment at the scale OpenAI predicts, that link has to break.
3. Training Data Compensation. Zero words about training data in an industrial policy document. The music industry solved a structurally identical problem a century ago with collective licensing (ASCAP, BMI). The same model works for training data. The only difference is that songwriters had the political power to demand it.
4. Compute as Public Utility. Data centers are industrial facilities. They should be governed like industrial facilities: county zoning, environmental review, community benefit agreements, revenue sharing. The Tennessee Valley Authority, the Bonneville Power Administration, Chattanooga's municipal broadband. The precedents exist.
5. Tax Base Modernization. OpenAI proposes "taxes related to automated labor" without naming rates, brackets, or mechanisms. We name them: an automation tax on AI-displaced labor, a compute tax, a data extraction fee, and international coordination through the OECD framework to prevent a race to the bottom.
6. AI-Enabled Direct Democracy. The biggest swing. A staged pathway from AI delegates for representatives (Stage 1, doable tomorrow) through citizen voice and informed direct democracy (Stage 6, maybe a decade out). Each stage has human checkpoints. The bandwidth argument against direct democracy evaporates with AI. We call it the Collapsium Proposal.
The framing problem
The paper also addresses what we see as a framing issue in OpenAI's document. There's a difference between "work with us to build the future" and "regulate us to protect the public." The auditing regime OpenAI proposes would apply only to frontier models, conveniently creating a regulatory moat around companies that can afford compliance. Nowhere in 13 pages does the word "antitrust" appear. The Progressive Era that OpenAI cites as precedent was defined by breaking up concentrated economic power, not by inviting the monopolists to help write the rules.
Read the full paper:
OpenAI asked for ambitious ideas. We sent ours to newindustrialpolicy@openai.com. If you have your own, they're listening.
