OpenAI — Industrial Policy for the Intelligence Age

Industrial Policy for the Intelligence Age is a policy document published by OpenAI in April 2026, framed under the heading "Ideas to Keep People First." It sets out the company's proposals for governing the transition toward what it calls superintelligence, organized around four pillars: sharing prosperity broadly, mitigating risks, democratizing access, and building infrastructure. The document's own two-part structure divides the proposals into "Building an Open Economy" and "Building a Resilient Society."

Summary of argument

The document presents OpenAI's position on how the transition to advanced AI should be governed. It forecasts a shift in AI capability from completing hours-long tasks to carrying out months-long projects, and describes a future in which, in its words, "superintelligence will speed up scientific and medical breakthroughs, significantly increase productivity, lower costs for families." On that basis it argues for an industrial-policy approach to the period it labels the intelligence age.

The document defines superintelligence as "AI systems capable of outperforming the smartest humans even when they are assisted by AI," states that "no one knows exactly how this transition will unfold," and positions itself as an opening contribution rather than a settled program: the ideas are described as "intentionally early and exploratory, offered not as a comprehensive or final set of recommendations." It focuses on the United States "as a starting point" while stating that the solutions must ultimately be global.

Four pillars

The proposal is structured around four pillars:

1. Share prosperity broadly. Everyone should see material improvements from AI; the document states that if AI winds up controlled by and benefiting only a few, "we will have failed."
2. Mitigate risks. The risks named are economic disruption, misuse in cybersecurity and biology, and loss of alignment or control. The document states that "as capability scales, safety must scale with it."
3. Democratize access. Broad participation should not depend on access to the most powerful models but on AI that is, in the document's terms, "affordable, privacy-preserving." It acknowledges that "some systems may need to be controlled for safety."
4. Build infrastructure. The document calls for large-scale investment in compute, energy, and data centers.

The framing chapter names five risks it says the company is "clear-eyed" about: jobs and entire industries being disrupted; bad actors misusing the technology; misaligned systems evading human control; governments or institutions deploying AI in ways that undermine democratic values; and power and wealth becoming more concentrated instead of more widely shared.

The case for industrial policy

The document draws an explicit historical analogy: the Progressive Era and the New Deal are described as having "modernized the social contract for a world reshaped by electricity, the combustion engine, and mass production," through labor protections, safety standards, social safety nets, and expanded access to education. It argues that the transition to superintelligence "will require an even more ambitious form of industrial policy."

The stated limiting principle is that markets are presumptively adequate — "in normal times, the case for letting markets work on their own is strong" — and that industrial policy has a role only "when market forces alone aren't sufficient." The proposed instruments are the government's existing toolbox of research funding, workforce development, market-shaping tools, and targeted regulation, sequenced so that nongovernmental institutions pilot and measure approaches first and governments then reinforce successes through procurement, regulation and investment. The document states that this public–private sequencing is intended to "stave off regulatory capture and centralized control."

Two near-term positions are stated directly: that "AI data centers should pay their own way on energy so that households aren't subsidizing them" and should generate local jobs and tax revenue; and that governments should implement "common-sense AI regulation—not to entrench incumbents through regulatory capture but to protect children, mitigate national security risks, and encourage innovation." The document also acknowledges a risk to its own author's position, stating that economic gains may "concentrate within a small number of firms like OpenAI."

Building an open economy

The first section sets out ten proposals addressed to labor-market disruption and the distribution of AI-driven growth.

• Worker perspectives. A formal mechanism for workers to collaborate with management on AI deployment, allowing workers to prioritize deployments that eliminate dangerous, repetitive, administrative or exhausting tasks, and to set limits on uses that intensify workloads, narrow autonomy, or undermine fair scheduling and pay.
• AI-first entrepreneurs. Microgrants or revenue-based financing paired with "startup-in-a-box" supports — model contracts, shared back-office infrastructure — so that workers can convert domain expertise into new companies; worker organizations are proposed as the delivery vehicle for training, shared services and commercial-terms negotiation.
• Right to AI. Treating access to AI as foundational to economic participation, by analogy to literacy campaigns and to electricity and internet buildout. The proposal is to expand affordable, reliable access to foundational models and make "a baseline level of capability broadly available, including through free or low-cost access points," with supporting education, infrastructure, connectivity and training. The document notes that the internet "still isn't fairly deployed across the globe or even the US" and proposes learning from that.
• Modernize the tax base. The stated concern is that AI may expand corporate profits and capital gains while reducing reliance on labor income and payroll taxes, eroding the base that funds Social Security, Medicaid, SNAP and housing assistance. The proposed response is greater reliance on capital-based revenues — higher taxes on capital gains at the top, corporate income, or targeted measures on sustained AI-driven returns — plus exploration of "taxes related to automated labor," paired with wage-linked incentives modeled on R&D credits that encourage firms to retain and retrain workers.
• Public Wealth Fund. A fund giving every citizen a stake in AI-driven growth, seeded jointly by policymakers and AI companies, investing in diversified long-term assets spanning both AI companies and firms adopting AI, with returns distributed directly to citizens.
• Accelerate grid expansion. Public–private partnership models to finance high-voltage interstate and interregional transmission, addressing financing constraints, permitting delays and siting risk. The named mechanisms are investment credits, flexible subsidies or equity stakes to reduce the cost of capital; removal of market barriers to advanced conductors and high-voltage direct current; and "a narrow federal authority to accelerate the construction of interregional transmission when it is in the national interest." Partnerships are to be structured to minimize taxpayer exposure and to translate into lower household and business energy costs.
• Efficiency dividends. Converting AI efficiency gains into worker benefits — larger retirement matches, a greater employer share of healthcare costs, child and eldercare subsidies — and time-bound 32-hour four-day-workweek pilots with no loss in pay, holding output and service levels constant, with reclaimed hours converted into a permanent shorter week or bankable paid time off.
• Adaptive safety nets. A three-step design: ensure unemployment insurance, SNAP, Social Security, Medicaid and Medicare are "fully functional, accessible, and responsive"; invest in real-time public measurement of AI's effect on work, wages, job quality and sectoral dynamics; then define a package of temporary expanded supports — flexible unemployment benefits, fast cash assistance, wage insurance, training vouchers — that "activates automatically when these metrics exceed pre-defined thresholds" and phases out as conditions stabilize. The document states this is designed to avoid a permanent expansion of programs.
• Portable benefits. Healthcare, retirement savings and skills training carried in accounts "attached to the individual, not the job," pooling contributions from multiple sources, with pooled retirement structures allowing continuous accrual across employers.
• Pathways into human-centered work. Expansion of the care and connection economy — childcare, eldercare, education, healthcare, community services — as a destination for displaced workers, supported by training pipelines and incentives to raise pay and conditions, complemented by a family benefit that treats caregiving as economically valuable and remains compatible with part-time work, retraining or entrepreneurship.
• Accelerate scientific discovery. A distributed network of AI-enabled laboratories integrating AI into experimental workflows, plus the physical infrastructure to translate validated discoveries into use, with the document specifying that both laboratory and production infrastructure "should be deployed broadly across universities, community colleges, hospitals, and regional research hubs, not concentrated in a small number of elite institutions."

Building a resilient society

The second section argues that safety work has concentrated on upstream safeguards — global standards, transparency around evaluations and mitigations, model testing, red teaming and usage policies — codified by policymakers "in the EU AI Act and in US state-based regulation," and that these efforts "should continue." It then argues that resilience will increasingly depend on what happens after deployment, when systems are monitored in real time and integrated into institutions not designed for agentic workflows. The analogies offered are electrical safety standards, automobile safety systems, aviation monitoring and coordinated response, and post-market surveillance in food and medicine — each described as built "with the luxury of time," which the document says this transition will not have.

Nine proposals follow:

• Safety systems for emerging risks. Tools to protect models, detect risks and prevent misuse across cyber, biological and other large-scale-harm pathways; use of advanced AI for threat modeling, red teaming, net assessments and robustness testing; complementary protective systems such as rapid production of medical countermeasures and expanded strategic stockpiles; and the creation of "competitive safety markets" through procurement, standards, insurance frameworks and advance-purchase commitments.
• AI trust stack. Provenance and verification standards, secure verifiable signatures for actions such as generating content or issuing instructions, and privacy-preserving logging and audit systems "capable of supporting investigation and accountability without enabling pervasive surveillance," alongside governance frameworks assigning responsibility to specific organizational roles.
• Auditing regimes. Strengthening the Center for AI Standards and Innovation (CAISI) to develop auditing standards for frontier AI risks in coordination with national security agencies, and using procurement, advance-purchase commitments, insurance frameworks and standards-setting to scale a competitive market of auditors. The document proposes that a narrow set of highly capable models — "particularly those that could materially advance chemical, biological, radiological, nuclear, or cyber risks" — may eventually require pre- and post-deployment audits, and states that such requirements should "apply only to a small number of companies and the most advanced models," with standards designed for international adoption to reduce fragmentation and compliance burden on small and cross-jurisdictional companies.
• Model-containment playbooks. Coordinated playbooks for containing dangerous systems already released, addressed to scenarios where "model weights have been released, developers are unwilling or unable to limit access to dangerous capabilities, or the systems are autonomous and capable of replicating themselves," on the argument from cybersecurity and public health that coordinated action reduces impact even where full containment is impossible.
• Mission-aligned corporate governance. Public Benefit Corporation structures with mission-aligned governance and explicit commitments to broadly shared benefits including long-term philanthropic giving, plus hardening frontier systems against corporate or insider capture by securing model weights and training infrastructure, "auditing models for manipulative behaviors or hidden loyalties," and monitoring high-risk deployments.
• Guardrails for government use. Rules codified in law and reinforced technically, with high standards for reliability, alignment and safety; use of AI-enabled auditing tools by inspectors general, congressional committees and courts; and modernization of transparency frameworks including the Freedom of Information Act, with clarification of "when AI-interaction logs and agentic action logs constitute federal records that must be retained."
• Mechanisms for public input. Published model specifications describing intended behavior, shared evaluation information, and representative public-input processes, so that "alignment isn't defined only by engineers or executives behind closed doors."
• Incident reporting. A mechanism for companies to report incidents, misuse and near-misses to a designated public authority, emphasizing "learning and prevention over punishment," with scoped public disclosure. Near-miss reporting is proposed to cover "cases where models exhibited concerning internal reasoning, unexpected capabilities, or other warning signals—even if safeguards ultimately prevented harm."
• International information-sharing. Expanding CAISI as a trusted technical evaluation body, then building "a global network of AI Institutes" with shared protocols for information exchange, joint evaluations and coordinated mitigation, potentially evolving into a multilateral safety-and-standards framework with secure cross-lab and cross-country channels and crisis communication. The document asks policymakers to enable safety- and risk-related information sharing "without running afoul of antitrust or competition constraints, using clear safe harbors and narrowly scoped information-sharing rules," and states the system should extend beyond national security to youth safety and well-being.

Follow-through

The closing section states three commitments: collecting feedback through a dedicated address (newindustrialpolicy@openai.com); a pilot program of fellowships and focused research grants of up to $100,000 and up to $1 million in API credits for work building on these ideas; and convening discussions at an OpenAI Workshop opening in May 2026 in Washington, DC.

Provenance

The document was published by OpenAI in April 2026 and sets out the company's policy positions; it is recorded on OpenAI. It positions the United States as needing industrial policy to compete globally, a framing connected to AI Race Dynamics, and presents an industry vision that runs alongside the government strategy described in America's AI Action Plan. The document assumes continued progress toward superintelligence, a premise linked to Recursive Self-Improvement, and explicitly acknowledges "jobs and entire industries being disrupted," connecting it to AI Labor Disruption.

As a company policy document it states positions rather than findings, and several of its proposals — targeted audits confined to "a small number of companies and the most advanced models," standards designed for international adoption to reduce fragmentation, and preemption-adjacent framing of state regulation as part of the existing upstream landscape — bear on questions where OpenAI is an interested party.

Relationships

• related: OpenAI — the publishing company and an interested party in several proposals.
• related: America's AI Action Plan — the government strategy alongside which this industry vision is offered.
• related: AI Labor Disruption — the open-economy section's adaptive safety nets, portable benefits and efficiency dividends are addressed to it.
• related: OpenAI Preparedness Framework V.2 — the upstream safeguard regime this document says should continue.
• related: NIST CAISI (Center for AI Standards and Innovation) — named as the institution to be strengthened for auditing standards and international coordination.
• related: EU AI Act (Regulation 2024/1689) — cited as existing codification of upstream safeguards.
• related: AI Pre-Release Vetting — the upstream evaluation regime the document says should continue, and the counterpart to its post-deployment proposals.