A Single Race Becomes a National Test Case
We have watched the artificial intelligence industry spend years lobbying quietly in the corridors of Washington. This week the fight broke into the open, and it did so over an unlikely figure: Alex Bores, a 35 year old New York state assemblyman and former Palantir engineer running in a Democratic primary that closes on June 23. According to Federal Election Commission filings reviewed by NPR, groups linked to OpenAI and Anthropic have collectively spent more than 15 million dollars on messaging for and against him. That is an extraordinary sum to aim at one junior candidate, and it tells us how seriously both companies treat the people who will write the next decade of AI law.
Bores earned the industry's attention by co-sponsoring New York's RAISE Act, a measure that would impose safety and transparency obligations on the largest AI developers. To OpenAI's allies he represents the kind of state-by-state regulation they want to stop. To Anthropic's allies he represents exactly the local guardrail they believe should survive. The result is a proxy war in which a congressional seat has become a referendum on whether AI should be governed from Washington alone or from fifty state capitals at once.
Inside OpenAI's War Chest
The OpenAI-aligned effort runs through a super PAC network called Leading the Future, funded by the venture firm Andreessen Horowitz, an OpenAI investor, and by OpenAI president and co-founder Greg Brockman. The network has raised more than 75 million dollars and has already spent 23.5 million across dozens of races stretching from Texas and Georgia to Illinois and Montana, channeling money through affiliated PACs including Think Big and American Mission. Its stated mission is to oppose policies that, in its words, stifle innovation or enable China to gain global AI superiority.
That framing matters because it bundles two arguments that play well in a midterm: economic competitiveness and national security. The practical goal, however, is narrower. OpenAI's camp wants Congress to set a single national approach to AI standards rather than allowing a patchwork of state laws to accumulate. By pouring money into primaries now, the network is trying to shape the composition of the next Congress before any federal framework is actually drafted, betting that friendly lawmakers will be easier to win over than hostile ones.
Anthropic Funds the Counterattack
Anthropic, founded by former OpenAI staff and long the louder advocate for regulation, is not sitting this out. It has backed super PACs countering the OpenAI-aligned assault on Bores, spending roughly 16.6 million dollars through vehicles such as Jobs and Democracy and Defending Our Values across North Carolina, Texas and Utah. In February the company committed 20 million dollars to a related nonprofit, Public First Action, whose mission is to oppose federal efforts that would freeze state progress without adequate federal safeguards.
The substance of Anthropic's position is the mirror image of OpenAI's. Where one camp wants preemption from Washington, the other wants to preserve the ability of states to move first. We read this not as a simple safety-versus-growth story but as a structural disagreement about where authority should sit. Both companies say they support some federal rules; they differ sharply on whether those rules should override the states. The dollars flowing into a New York primary are the clearest evidence yet that the disagreement is now worth real money to both sides.
What the Money Actually Buys
Skeptics rightly ask whether 15 million dollars can change the outcome of a single primary, and the honest answer is that nobody knows. Molly White, an independent researcher who tracks the technology industry, cautioned that even a Bores victory would be modest in isolation. As she put it, if Alex Bores is elected, one fairly junior congressperson is not likely to have an enormous impact on the ultimate AI regulation that might be passed in the next couple of years. The spending is less about one vote than about deterrence and signaling to every other candidate watching.
Michael Beckel, director of money in politics reform at Issue One, framed the stakes more broadly. This type of spending, he said, really helps shape who is at the table and what perspectives they are bringing into those conversations when new legislation is crafted. Katie Harbath, a former Facebook public policy official who now runs the firm Anchor Change, called the moment a real experiment this time to see if that sort of money can really sway any of these races in the way that these companies want it to. The midterms, in other words, have become a live test of AI's political muscle.
A Policy Vacuum the Industry Is Racing to Fill
Part of what makes the spending so aggressive is that the underlying rules remain unwritten. Adam Kovacevich, founder and chief executive of the industry group Chamber of Progress, observed that there seems to be little energy being spent on actually writing those standards before this year's midterm elections. That vacuum is precisely what both PAC networks are trying to exploit. Money spent now buys influence over who gets to fill the blank page later, which is a far better return than lobbying against a bill that already exists.
Nicole Alvarez, a senior technology policy analyst at the Center for American Progress, distilled the conflict to a single line: the real fight is what governance looks like. For context, in 2025 OpenAI, Meta, Alphabet and Nvidia together spent 50.9 million dollars lobbying Congress. The midterm super PAC surge is a continuation of that campaign by other means, moving the contest from committee rooms to primary ballots where the cost of entry is lower and the leverage is higher.
Why CIOs and CTOs Should Be Watching
It would be easy for enterprise leaders to dismiss this as Washington theater, but we would urge the opposite. The outcome of the preemption debate will determine whether your compliance teams answer to one federal regime or to a growing thicket of state laws, each with its own disclosure, audit and liability requirements. A national standard would simplify deployment; a state patchwork would raise the cost of operating AI systems across markets. Either way, the rules being fought over today will land on enterprise roadmaps tomorrow.
The deeper lesson is that the companies supplying your models are now political actors with competing regulatory agendas, and those agendas are not neutral. When a vendor argues for or against a given rule, it is also arguing for the competitive landscape that suits it best. We think technology executives should factor that into procurement and governance decisions, reading vendor policy positions as carefully as they read service-level agreements. The midterms will not settle AI regulation, but they will tell us who holds the pen when it finally gets written.
