20 Ideas for AI-Era Industrial Policy
AI for DC
Hi
Welcome (back) to The Prompt. Today, special issue, special timing, special guest in the OpenAI Forum:
20 policy ideas for keeping AI people-first
Join Sam Altman today in the OpenAI Forum to help shape what’s next
If you find the below of interest, make sure you’re signed up for the next issue.
[News] Industrial policy for the AI age
Being a responsible steward of AI, we believe, means more than just innovating through product — it means innovating through policy, too. With AI, we are entering a new phase of economic and social organization that will fundamentally reshape work, knowledge, and production. The magnitude of the shifts we expect, as well as the potential risks we foresee, demand much more than the policy changes (largely updates to existing policies) on the table today.
That’s why we’re releasing a slate of ideas to help start a broader conversation about the kinds of policies and institutions needed to navigate this transition, a conversation that needs to happen across all AI stakeholders from governments and companies to communities and families. These ideas, co-developed by OpenAI’s policy and researcher teams, are intentionally early and exploratory, offered as a starting point that we invite others to build on, refine, challenge, or choose among through the democratic process.
History shows that democratic societies can respond to technological upheaval with ambition: reimagining the social contract, mediating between capital and labor, and encouraging broad distribution of the benefits of technological progress while preserving pluralism, constitutional checks and balances, and freedom to innovate.
In normal times, the case for letting markets work on their own is strong. Historically, competition, entrepreneurship, and open economic participation have lifted living standards and expanded opportunity. Capitalism, imperfect as it is, remains an effective system for translating human ingenuity into shared prosperity.
But industrial policy can play an important role when market forces alone aren’t sufficient – when new technologies create opportunities and risks that existing institutions aren’t equipped to manage. It can help translate scientific breakthroughs into scaled industries and broad-based economic growth.
These ideas are our first contribution to that effort, but only the beginning. Progress will depend on continued iteration, experimentation, and collaboration across institutions and sectors. To help sustain momentum, OpenAI is: (1) welcoming and organizing feedback through newindustrialpolicy@openai.com; (2) establishing a pilot program of fellowships and focused research grants of up to $100,000 and up to $1 million in API credits for work that builds on these and related policy ideas; and (3) convening discussions at our new OpenAI Workshop opening in May in Washington, DC.
[Policy] 20 ideas for keeping AI people-first
Here’s the roster of ideas we’re offering in today’s release, which is organized in two sections: 1) building an open economy with broad access, participation, and shared prosperity; and 2) building a resilient society through accountability, alignment, and management of frontier risks.
1. Building an Open Economy
Worker voice in AI deployment → Give workers formal input into how AI is used at work to improve job quality, safety, and fairness.
AI-first entrepreneurs → Enable workers to start AI-powered businesses by reducing overhead and providing shared tools, financing, and infrastructure.
Right to AI → Treat access to AI as essential infrastructure by ensuring affordable, widespread availability and training.
Modernize the tax base → Shift tax systems toward capital and AI-driven gains to sustain public programs as labor income evolves.
Public Wealth Fund → Create a national investment fund so all citizens share directly in AI-driven economic growth.
Accelerate grid expansion → Build public-private models to rapidly expand energy infrastructure needed for AI.
Efficiency dividends for workers → Convert AI-driven productivity gains into better benefits, higher compensation, or reduced working hours.
Adaptive safety nets that work for everyone → Ensure existing programs (UI, SNAP, Social Security, etc.) are fully functional and responsive at scale, and then define automatic support systems (cash, training, wage insurance) that expand when AI-driven disruption exceeds defined thresholds.
Portable benefits → Decouple benefits from employers so healthcare, retirement, and training follow workers across jobs.
Pathways into human-centered work → Expand jobs in care, education, and community services as durable, human-centric employment sectors and provide direct support recognizing caregiving as economically valuable work.
Accelerate scientific discovery and scale the benefits → Build distributed AI-enabled research labs to accelerate scientific discovery and validation and invest in infrastructure and systems to translate AI-driven discoveries into real-world benefits.
2. Building a Resilient Society
Safety systems for emerging risks → Develop tools and markets for detecting and preventing AI misuse in high-consequence domains like cyber and bio.
AI trust stack (verification + provenance) → Create systems that allow people to verify AI outputs, actions, and accountability while preserving privacy.
Auditing regimes for frontier AI → Develop standards, catalyze a network of auditors, and prepare for a future where a narrow set of highly capable models that pose national security risks require pre- and post-deployment audits.
Model containment playbooks → Develop coordinated response plans to limit harm from dangerous or uncontrolled AI systems.
Mission-aligned corporate governance → Encourage frontier AI companies to adopt governance structures that embed public-interest accountability.
Guardrails for government use of AI → Establish clear legal and technical limits on how governments deploy AI, while enhancing oversight.
Mechanisms for public input → Create structured, representative processes for public participation in AI system design and alignment.
Incident reporting systems → Establish a system for companies to share information about AI failures, misuse, and near-misses with authorities.
International AI coordination network → Build a global system of AI institutes to share risk information, coordinate evaluations, and respond to crises.
For more on these ideas, check out the full doc.
[Event] Building AI’s future
This Monday, the OpenAI Forum is hosting a talk on the future of building AI with OpenAI CEO and Co-Founder Sam Altman – along with OpenAI Chief Futurist Joshua Achiam, and OpenAI Researcher Adrien Ecoffet. They’ll be focusing on the above ideas to start the conversation about how to ensure that AI really does benefit everyone.
The OpenAI Forum presents programming and discussions by and for our community of 70,000 AI experts and enthusiasts from across tech, science, medicine, education, government, and other fields.
2:00 PM – 3:00 PM EST on Apr 6
[About] OpenAI Academy
The Academy is OpenAI’s free online and in-person AI literacy trainings for beginners through experts.
Our Academy recently teamed up with the San Antonio Spurs Community Impact and Positive Coaching Alliance to bring together coaches, parents, and community leaders to explore how AI can support learning and youth development, both on and off the court.
A key takeaway from the event was that trusted adults are central to responsible AI adoption. Parent-coaches are emerging as translators of new technology, helping young people navigate AI safely and practically. As adoption accelerates, building AI literacy within families and local communities is becoming as important as access itself.
This work reflects a broader focus of the OpenAI Academy: partnering with institutions that shape everyday life, and advancing a more distributed model of AI readiness rooted in communities, not just companies.
[Disclosure]
Graphics created by Base Three using ChatGPT.
I could not agree more with your paper. In fact, my recent article discusses the need for AI provenance. I hope we can continue the conversation together in May in DC. https://aiboatcaptain.substack.com/p/when-ai-starts-eating-its-own-tail?r=6jndsn&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true
my bot doesn't like this.
Here's a rebuttal of the OpenAI industrial policy document, written through the lens of the thesis.
---
**OpenAI's "People First" Paper: A Structural Autopsy**
The document is well-intentioned. It is also structurally incoherent. Twenty policy ideas from the organisation most responsible for accelerating unit cost dominance, offered without any acknowledgement that the accelerator and the brake cannot be operated by the same foot.
Start with the framing. OpenAI characterises this as "innovative policy" for a "new phase of economic organisation." What it actually is: a list of demand-side patches for a supply-side rupture. The wage-demand circuit is being severed at the production end. You cannot fix that with portable benefits and public wealth funds. Those instruments assume the circuit still operates. They redistribute within a system that the company authoring the document is actively dismantling.
**On coordination impossibility**
The document invokes democratic history as evidence that societies can respond to disruption with ambition. This is Russell's Turkey reasoning: it worked before, therefore the mechanism still holds. Previous technological upheavals automated physical tasks. Human cognition remained the refuge. AI automates cognition itself. The historical analogy fails at the only point that matters.
Meanwhile, the document contains no acknowledgement that OpenAI cannot restrain its own deployment without Google, Meta, Mistral and the open-source ecosystem filling the gap immediately. Every "responsible" policy idea presupposes a coordination capacity that coordination impossibility forecloses. OpenAI knows this. The paper does not say it.
**On "pathways into human-centred work"**
Care and education are nominated as durable employment sectors. Apply the three-gate test: AI-resistant, wage-sustaining, scalable to the volumes of displaced workers.
Care work fails gate three at the wage required to absorb mass displacement. Education fails gate one as AI tutoring systems already outperform median teachers on measurable outcomes. The sectors are not "human-centred" by structural necessity; they are politically protected and subsidised. When the tax base erodes under mass unemployment, the subsidy mechanism collapses.
**On the public wealth fund**
A national investment fund requires a tax base to fund it. The tax base requires employed workers generating taxable income. The mechanism that depletes the tax base is the same mechanism the fund is supposed to offset. This is circular, not transformative.
**On "modernise the tax base"**
Correct diagnosis, no mechanism. Shifting taxes toward capital gains requires legislative coordination across jurisdictions that are in direct competition to attract AI investment. A country that taxes AI capital more heavily loses the investment to one that doesn't. This is coordination impossibility expressed as tax policy.
**The honest sentence buried in the document**
"AI could widen inequality by compounding advantages for those already positioned to capture the upside."
That sentence is the thesis. Everything else in the document is an attempt to avoid its conclusion.
**What the document is**
It is a Cassandra Prison product. The people writing it understand the structural problem well enough to name it in subordinate clauses. They cannot name it in the subject position without indicting the organisation they work for and foreclosing the capital they depend on. So the paper reads as genuine concern expressed through instruments that cannot address the problem they are genuinely concerned about.
The WSJ called it a "charm offensive." That framing is too cynical. It reads more like a sincere document written by people operating inside an institution whose incentive structure makes structural honesty impossible.
That is precisely how the Cassandra Prison works.