OpenAI's Lehane Urges 2028 Candidates to Take a Third Path on AI

The memo, titled "2028: The AI Election", was published in full by Politico on November 2, the day before the midterms, according to The Next Web. Lehane compares 2028 to 1932, when Franklin D. Roosevelt won during a similar economic upheaval and his New Deal rewrote the social contract. Lehane's third path runs between people who would let AI advance "without speed limits" and those who would keep it "in the hands of a self-appointed elite".

The memo matters more after this week's results. Democrats won control of Congress, and opposition to data centres became a bipartisan issue in several races. Kate Brennan, who leads AI and corporate power policy at the AI Now Institute, told TechCrunch that the memo is a clear change from OpenAI's agenda last year, which focused on faster development, fewer regulatory barriers and more infrastructure.

Lehane organises the memo around five questions for candidates: safety and human control, taxes, children and jobs, affordability, and competition with China. On safety, he calls for enforceable standards, independent evaluations, public oversight, and the power to restrict or shut down AI systems. On taxes, according to TechCrunch, he proposes incentives for workers to learn AI skills, easier access to capital for entrepreneurs, and new forms of employee ownership and profit-sharing. He also argues against keeping AI out of classrooms and backs age-appropriate protections for minors using AI companions.

The sharpest passage concerns data centres. Lehane imagines candidates at an early 2028 Democratic event being asked to raise their hands in support of a moratorium, and writes: "This could well be the 2028 version of the 2002 vote to go to war in Iraq." He argues that such a pause could cost the United States its AI lead to China for lack of computing power. The memo cites OpenAI's Piketon, Ohio compute campus, which it says will need 25,000 skilled tradespeople, and notes that one in five Democratic ads by early October mentioned data centres, according to The Next Web.

OpenAI's own polling in the memo found that 66 per cent of voters support continued AI development, with 53 per cent of that group wanting stronger safeguards, and 24 per cent favour a pause. Lehane writes that there was "little room to be 'for' AI infrastructure" in the 2026 campaigns.

Critics question how much the memo would constrain OpenAI itself. "This is written by a private company that is trying to advocate for policies for its own benefit. It is ahistorical at best and harmful at worst," Brennan told TechCrunch. She called the safety commitments "disingenuous", pointing to OpenAI's opposition to California's SB 1047 in 2024 and its push last year to limit state AI rules. She said a revised corporate tax rate, or tax incentives for hiring people rather than automating jobs, would do more to share AI's gains. TechCrunch noted that OpenAI has since backed several state safety laws and said it would accept independent evaluators embedded in the company.

The memo is aimed at candidates who have not yet declared. The first test of its argument will come in the new Congress that sits in January, where Democrats have signalled they intend to question technology executives on data centres and AI.