OpenAI Sets Out Its AI Policy Agenda for the 2028 Election
OpenAI's Chris Lehane pitches a five-part AI agenda for 2028, as critics question worker gains, data centers, safety rules and corporate accountability.
Summary
OpenAI chief global affairs officer Chris Lehane is positioning AI for the 2028 presidential race, arguing in a November 4, 2026 essay for a third path between uncritical adoption and anti-AI backlash. Comparing AI disruption with Franklin D. Roosevelt’s 1930s New Deal, he proposes five priorities spanning worker protections, education, AI safety and data centers, with child safeguards and human control also central. Democrats just flipped Congress amid anger over inflation, tariffs and living costs, while bipartisan data center opposition grew in several races. Voters fear higher electricity bills, job replacement and gains concentrated among wealthy companies.
Lehane proposes tax incentives for AI training, easier capital for entrepreneurs, and employee ownership and profit-sharing approaches, but no binding wealth-sharing or worker-displacement costs. He emphasizes data center construction jobs and local tax revenue, warns moratorium supporters could surrender the US lead to China, and compares that choice with Congress’s 2002 Iraq war authorization. He supports age-appropriate protections for minors and AI companions, while arguing classroom restrictions could disadvantage rural students despite unproven learning gains.
Lehane also calls for enforceable safety standards, independent evaluations, public oversight, kill switches and human authority over consequential decisions. Yet studies find OpenAI and other labs lack comprehensive loss-of-control plans. OpenAI opposed California’s SB 1047 in 2024 and sought limits on state regulation in 2025, before recently backing several state laws and embedded independent evaluators. AI Now Institute policy lead Kate Brennan says the agenda assumes mass adoption, benefits OpenAI and lacks corporate accountability; she favors revised corporate taxes, incentives for hiring humans and market-power limits. OpenAI provided no additional specifics before publication.
Positives
- Five policy priorities bring worker protections, education, AI safety and data center development into the 2028 policy debate.
- Enforceable safety standards, independent evaluations, public oversight and kill switches would create stronger mechanisms for controlling dangerous AI systems.
- Age-appropriate protections and safeguards for minors using AI companions make child safety a central policy concern.
- Tax incentives for AI training, employee ownership and profit sharing could broaden access to AI-related economic gains.
- OpenAI now backs several state safety laws and embedded independent evaluators after previously resisting comparable regulation.
Risks & concerns
- Lehane proposes no binding requirements for sharing AI-generated wealth or making companies pay worker-displacement costs.
- Data center expansion could increase electricity bills even as OpenAI emphasizes construction jobs, local taxes and competition with China.
- Classroom AI access is promoted as an equality measure although its ability to improve learning remains unproven.
- OpenAI and other laboratories have not published comprehensive plans for responding if advanced models escape human control.
- OpenAI opposed California’s SB 1047 in 2024 and sought to limit state regulation in 2025, weakening its current safety message.
- Kate Brennan says the agenda assumes mass adoption while failing to constrain corporate power or market concentration.