Insider Brief
- OpenAI’s policy paper argues that governments must adopt a new industrial policy to manage AI’s economic disruption while ensuring its benefits are broadly shared.
- The proposal outlines measures to build an open AI economy, including worker participation, expanded access to AI tools, tax reforms, public wealth funds, modernized safety nets, and investment in infrastructure and human-centered jobs.
- It also calls for a resilient governance framework with stronger safety systems, auditing regimes, accountability standards, and international coordination to manage risks from increasingly powerful AI systems.
OpenAI is urging governments to adopt a sweeping new industrial policy to manage the economic and social disruption expected from advanced artificial intelligence, arguing that current institutions are not equipped for the transition.
In its April 2026 paper, Industrial Policy for the Intelligence Age: Ideas to Keep People First, the company outlines a framework for steering the development of increasingly capable AI systems — what it describes as a move toward “superintelligence” — in a way that preserves broad economic participation, democratic control and social stability.
The document presents a clear warning that without intervention, AI could concentrate wealth and power while displacing workers and straining existing safety nets. At the same time, it frames the technology as a potential driver of faster scientific discovery, lower costs for essential goods and higher overall living standards.
The proposed solution is not incremental regulation but a rethinking of industrial policy that combines public investment, market incentives and governance mechanisms to shape how AI integrates into the economy.
A Technological Shift With Economic Consequences
The paper situates AI alongside past general-purpose technologies such as electricity and mass production, arguing that each required new policy frameworks to distribute benefits and manage disruption.
AI’s rapid progress is already changing how work is done. Systems have advanced from assisting with short tasks to completing work that once took hours, and may soon handle projects that take months. That trajectory will reshape organizations, industries, and labor markets, according to the OpenAI analysts.
The paper acknowledges potential job displacement, misuse in areas such as cybersecurity and biology, the possibility of systems acting outside human intent, and the concentration of economic gains among a small number of firms.
The core policy challenge is ensuring that AI expands opportunity rather than narrowing it.
Building an Open Economy
The first pillar of the proposal focuses on broad participation in the AI economy.
A central recommendation is to give workers a structured role in how AI is deployed. The paper indicates that employees are best positioned to identify where automation can improve safety, reduce repetitive work and increase job quality and where it could instead erode autonomy or intensify workloads.
The analysts also propose lowering barriers to entrepreneurship. By using AI to handle administrative tasks such as accounting, marketing and procurement, workers could more easily launch businesses based on their domain expertise. The paper suggests pairing this with microgrants, shared services and standardized tools to help small firms compete.
Access to AI itself is framed as a foundational issue. The paper calls for a “right to AI,” likening it to past efforts to expand access to electricity and the internet. This would include affordable access to core AI systems, along with the infrastructure and training needed to use them effectively, particularly for underserved communities.
Tax policy is another focus. As AI shifts economic activity away from labor income and toward capital, the paper warns that existing tax systems could weaken. It proposes rebalancing the tax base toward capital gains, corporate income and potentially new forms of taxation tied to automated labor, while offering incentives for companies to invest in workers.
To distribute gains more directly, the paper introduces the idea of a Public Wealth Fund. Under this model, governments and AI companies would contribute to a fund that invests in AI-related growth, with returns distributed to citizens. The goal is to give individuals a direct stake in the economic upside of AI.
Infrastructure is also central. The paper calls for accelerated investment in energy systems to support AI data centers, including public-private partnerships to expand power grids while ensuring that households are not burdened with higher costs.
Another proposal focuses on translating productivity gains into tangible benefits. Companies could be incentivized to share efficiency gains through higher retirement contributions, better healthcare coverage, or reduced working hours, including experiments with a four-day workweek.
The paper also emphasizes strengthening and modernizing safety nets. It proposes systems that automatically expand support — such as unemployment benefits or wage insurance — when economic disruption reaches predefined thresholds, then scale back as conditions improve.
Longer-term, the analysts call for portable benefits that are not tied to a single employer, allowing workers to carry healthcare, retirement savings and training support across jobs and industries.
Finally, the paper identifies “human-centered” sectors such as healthcare, education and caregiving as key areas for job growth. While AI may assist in these fields, human interaction will remain essential, making them a potential destination for workers displaced by automation.
Building a Resilient Society
The second pillar focuses on managing the risks associated with more powerful AI systems.
The paper suggests that current efforts — such as model testing, safety evaluations, and usage policies — are necessary but not sufficient. As AI systems are deployed more widely, new risks will emerge in real-world conditions.
To address this, the authors propose developing “safety systems for emerging risks,” including tools to detect misuse in high-stakes domains like cybersecurity and biological research. They also call for using AI itself to model threats and test system robustness.
A key concept is the creation of an “AI trust stack”, which is a set of technologies and standards that allow users to verify AI-generated content and actions. This could include digital signatures, audit logs, and privacy-preserving monitoring systems to ensure accountability without enabling surveillance.
The paper also recommends building formal auditing regimes for advanced AI systems. These would involve independent evaluators assessing safety and security risks, with stricter requirements applied to the most capable models.
In extreme scenarios, where dangerous systems cannot be easily contained, the authors propose developing coordinated response plans — referred to as “model-containment playbooks” — to limit harm and manage the spread of capabilities.
Corporate governance is another area of focus. The paper suggests that leading AI companies adopt structures that embed public-interest obligations into decision-making, such as public benefit corporations, and implement safeguards against internal misuse of powerful systems.
Governments, too, are a focal point. The paper calls for clear rules governing how public agencies can use AI, with high standards for safety and reliability. At the same time, it suggests that AI could improve transparency by creating detailed records of decision-making processes that oversight bodies can review.
Public input is emphasized as a critical component of governance. According to the analysts, decisions about how AI systems behave should not be left solely to companies or engineers, but should include structured mechanisms for broader societal participation.
The paper also proposes incident-reporting systems that allow companies to share information about failures, misuse, and near-misses, with the goal of improving collective understanding and prevention.
Finally, it calls for international coordination. The authors envision a network of national AI institutes that share information about risks, evaluations, and mitigation strategies, potentially evolving into a global framework for AI governance.
The paper closes with a call for immediate engagement, arguing that the transition to advanced AI is already underway and that policy decisions made in the near term will shape outcomes for decades.
It presents its recommendations not as a final blueprint but as a starting point for broader discussion among governments, companies, and civil society.
