Campaigners backing tighter artificial intelligence regulation outspent industry allies in the November 2026 US midterm elections, as Democrats captured four Republican-held congressional seats targeted by the safety campaign. Opposition to local data centres helped turn an abstract technological argument into a contest over neighbourhoods, energy use and jobs.

Pro-regulation groups spent $44.4 million against $27.9 million from industry supporters. In Michigan, victorious Democrat Will Lawrence campaigned against local server farms. Exit polls indicated that AI and data centres were decisive or major considerations for 56% of voters, although that does not establish how many votes changed because of them.

The confrontation followed an unsuccessful attempt to prevent states from enforcing their own AI rules for a decade. A Politico investigation identified technology companies’ efforts to advance that provision through the Trump administration’s budget legislation in May 2025 as a turning point. The Senate rejected it by 99 votes to one that July.

In August 2025, Andreessen Horowitz and OpenAI established Leading the Future with $100 million, seeking to deter legislators from slowing deployment. The approach followed cryptocurrency interests’ use of campaign spending against regulatory opponents.

It also provoked a well-funded counterweight: technology-policy veterans, consumer advocates, competition campaigners and wealthy AI engineers concerned about the technology. Former congressman Brad Carson founded the Public First network; lawyer Jay Schuster organised Silicon Valley donors, alongside Democratic strategist Rishi Bharwani.

The division ran through the industry itself. Sam Altman’s OpenAI pressed for faster development and fewer barriers; Dario Amodei’s Anthropic backed regulation and channelled about $40 million to organisations linked to Public First ahead of its expected flotation.

Opponents of regulation argued that restrictions would obstruct progress and hand China technological leadership. Effective altruists and safety researchers warned instead of possible human extinction over decades or centuries. Campaigners brought that distant risk closer to voters through concerns about job losses, children’s online safety and noisy, energy-intensive data centres in residential areas.

Democratic election advertising mentioning data centres rose from none during 2020–24 to 659 adverts in this cycle. Another 387 adverts directly attacked technology giants; the combined advertising expenditure exceeded $103 million.

Money did not guarantee victory. Alex Bores, the safety camp’s leading New York candidate, lost June’s Democratic primary despite $19 million in supportive spending against $8 million from the opposing lobby. The broader result nevertheless demonstrated that politicians challenging AI companies could attract substantial financial backing. Electoral success is not the same as enacted regulation; the immediate change was in the political cost of resisting it.