OpenAI's Chris Lehane wants AI to define the 2028 election: five tests, one warning on data centers

OpenAI's top policy executive has published a plan to make AI the defining issue of the 2028 presidential race. Chris Lehane, OpenAI's chief global affairs officer, released an essay on November 4, 2026 about the 2028 election and AI. It names five tests for candidates, urges a "compelling third path" on AI, and warns that backing a data center moratorium could cost America its lead over China. It also marks a visible shift in how the company talks about regulation.

What the essay says

The essay appeared on OpenAI's global affairs Substack on November 4, after Democrats flipped control of Congress in the midterms, according to TechCrunch's Rebecca Bellan. Lehane argues that the 2028 election could be a defining moment for AI and that politicians must address public concerns. He compares AI's economic impact to the New Deal.

On the central choice, he writes that the candidate to watch is one "who can articulate a compelling third path between those who would let AI advance without speed limits" and the other camp the essay names. We quote only the part of that sentence we could verify word for word. He also writes: "AI is not a living being but a tool to expand what people can learn, create, build, and produce."

The essay lays out five priorities by which voters could judge candidates:

• Safety and human control. Enforceable safety standards, independent assessments, the ability to restrict or shut down systems, and a right to challenge consequential AI decisions.
• Tax and economic participation. Policy that spreads AI's gains through incentives for learning, access to capital, and broader ownership or profit-sharing.
• Kids and workers. Protections for minors from harmful AI interactions, expanded educational access, and support for the skilled trades that build AI infrastructure.
• Affordability and opportunity. Judging AI policy by whether it lowers costs and widens access to healthcare, education, government services and small business growth.
• Competitiveness with China. Whether America invested in energy, computing, research and talent to keep its lead.

The data center line

The most pointed passage concerns data centers. Lehane writes that "aspiring candidates could be asked to raise their hands to show support for a moratorium on data centers." TechCrunch reports he compares that stance to the 2002 Iraq war authorization vote, a vote candidates were later made to answer for. He stresses construction jobs and local tax revenue.

This is a direct answer to a fight already underway. The San Francisco Standard reported on October 31 that proposed data center moratoriums have appeared in nearly 500 communities over the last year. Our data center backlash tracker logs the specific bans, lawsuits and blocked projects.

A change in tone, and its critics

TechCrunch points out that OpenAI's position has moved. In 2024 the company opposed California's SB 1047, which would have required catastrophic-risk evaluations and shutdown capability. In 2025 it sought to limit state regulation, with a policy agenda focused on accelerating development. More recently, TechCrunch notes, it has backed several state safety laws and said it would accept independent evaluators embedded in the company.

Kate Brennan of the AI Now Institute told TechCrunch the essay marks a notable shift from the 2025 agenda. She argued the New Deal comparison invokes state capacity, which she said is not what a private company is pushing for, and called OpenAI's safety stances "disingenuous." She also noted that the tax and worker proposals are not binding requirements to share AI wealth or pay for displaced workers. TechCrunch asked OpenAI for specifics and had no response by publication.

Two more points from the reporting are worth stating. First, TechCrunch notes that recent studies found OpenAI and other labs have not published comprehensive plans for containing a rogue model, which sits awkwardly beside a call for the ability to "restrict or shut down" systems. Second, Lehane argues against keeping AI out of classrooms, while critics say the claim that AI improves learning is unproven.

How it connects to the midterm spending

Lehane's essay lands after an expensive cycle. OpenAI-aligned groups, led by the super PAC network Leading the Future, spent roughly $26 million in primaries and far less in the general election, per the San Francisco Standard. The newspaper reported that OpenAI president Greg Brockman, who gave $25 million last year, pulled out of a second planned $25 million donation in June, calling the PAC "a distraction." We track the money by group in a separate ledger of AI super PAC spending.

The essay can be read as the next step in a shift from spending to argument: if ads did not move voters, a framework candidates can adopt might. We cannot say that is OpenAI's intent, and the company has not said so.

We also covered OpenAI's earlier post-midterm push in our report on its proposed speed limits and compute campuses. The new essay is the public, longer version.

What we would watch

• Specifics. The essay sets tests, not bills. Watch for named legislation, tax language or a lobbying position that matches the "enforceable standards" promise.
• Consistency. OpenAI's record on SB 1047 and state preemption will be cited by opponents every time the company praises safety rules.
• Candidate uptake. If 2028 hopefuls adopt the "third path" phrasing, the essay worked as framing. If they ignore it, it stays a company document.

Our position: read the essay as a signal that the company expects AI to be a ballot issue and wants to set its terms. The five tests are reasonable. A company that writes the test for its own regulation should be judged by what it supports when a bill has a number.