Public First changes the AI Super PAC Landscape

A couple months ago, a few notables from the tech industry (including Andreessen Horowitz and OpenAI’s Greg Brockman) announced intentions to spend >$100M on super PACs, to intimidate politicians away from regulating AI. Their plan was to copy a strategy that the crypto industry used in 2024 to prevent regulation of crypto.

The basic strategy, which I described in a Twitter thread and which Scott Alexander described in an ACX post, is to use super PAC money to run political ads in favor of politicians who are good on your issue (and against those who are bad on your issue).

The point isn’t to persuade voters that your position on crypto/AI is right on its merits, but instead to set political incentives for elected politicians to side with your position, regardless of what voters want. Crucially, the political ads don’t mention crypto/AI, but instead focus on saying nice things about your preferred candidate (or mean things about your dispreferred candidate) on issues that voters care most about, like healthcare or inflation.

And the level of announced super PAC spending would make the Andreessen-OpenAI super PAC (known as “Leading the Future,” or LTF) one of the largest industry-based super PACs in America, likely second only to Fairshake, the crypto super PAC that LTF is copying. (The only larger super PACs tend to do things like support one party over the other or support a particular candidate for president.)

After analyzing the situation, I concluded that, “Unless something changes, we should expect the AI industry will achieve similar political dominance as crypto”, meaning “approximate political dominance”.

Well, something has indeed changed.

Last Tuesday, it was reported by the NYT that people who are worried about the direction of AI are setting up their own super PAC, “Public First”, as a counter to LTF. Public First has plans to spend $50M–$100M in the 2026 elections, and it’s being led by two former members of Congress (one D and one R) - Brad Carson & Chris Stewart.

Given what’s currently known about the competing efforts, Public First appears well-positioned to successfully push back against LTF, and the net effect of both efforts may even be to empower the pro-“regulate the AI industry” side.

There are several reasons why we should expect Public First to be able to punch above its weight and serve as an effective counter to LTF.

1. Declining marginal value of political spending

Reporting indicates Public First will spend a substantial fraction of what LTF spends. But political spending has declining marginal value, meaning Public First’s spending should be more than proportionately as effective, all else equal:

• Money in primary races goes further than money in general elections.

• In primaries, the first chunk of money spent (to gain name recognition for your candidate) goes further than subsequent chunks of money spent (to persuade voters who already know about your candidate that they should vote for them).

• Money in the races you care more about goes further for advancing your cause than money in races you care less about, as the races you care more about matter more.

• Etcetera

The amount of money being discussed here on both sides is so large that each side will likely pick up all the low hanging fruit available, and marginal money will be spent in more marginal ways.

2. The Public is firmly on Public First’s side

In basically every poll about AI politics, you find:

A) The public is skeptical of AI, and
B) The public wants more regulation of AI.

This is true whether the poll asks about high-level questions or hypothetical policies or actual bills, and it’s true whether you look at the responses of Democrats or Republicans.

The tech industry has tried to craft polls that show the opposite results. But in order to get the results they want, they’ve had to resort to some truly bonkers polling methodology. Which kinda proves the point in itself.

In short, the public is on Public First’s side.

Now, if you have a situation where voters are on one side of an issue and large political donors are firmly on the other side, the political calculus for candidates can be tricky. But if you have large political donors on both sides, with voters overwhelmingly on one side, then the math is easier – politicians will be incentivized to go to the side with the voters. Which in this case is Public First.

And if Public First is seen as “for the public” while LTF is seen as “against the public” (which, again, would agree with all the polling on the issue), then being supported by LTF might become a negative factor in itself, somewhat diminishing the effectiveness of LTF’s resources. Already, we’re seeing early signs that LTF’s money is becoming a little toxic, with LTF’s first targeted opponent, Alex Bores (sponsor of NY’s RAISE Act) actively campaigning on the fact that LTF is opposing him.

3. In their heart of hearts, most politicians probably side more with Public First

First, politicians are people, and by default their impulses on a random issue generally won’t be that different from the general public. Given where the public stands on AI, most politicians probably also have a somewhat similar impulse.

Second, for all the cynicism towards politics, politicians do generally want to be representing the interests of their constituents, all else equal, instead of voting against their constituents’ preferences.

Third, many politicians are kinda fed up with Big Tech. Or at least fed up with the part of Big Tech that constantly pushes for absolutely minimal regulations and no accountability for their industry.

At the same time, successful politicians generally won’t go around alienating powerful actors for no good reason. And Big Tech is powerful, including the parts that push a radical accelerationist agenda with no rules.

But Public First may now be giving these politicians a choice they didn’t have before: you could ally with the more extreme and less responsible parts of the tech industry, or you could ally with those of us who want to put guardrails on the technology. Either way you can find substantial support.

4. Spending on both sides increases the political salience of AI

When LTF spends money, it’ll be seen as an effort to reduce AI regulations, and thereby increase the salience of AI politically. When Public First spends money, it’ll be seen as an effort to increase AI regulations, and thereby increase the salience of AI politically. Even just the fact that there are now two very large super PACs on opposite sides of an issue is juicy and likely to increase salience.

And guess what increased salience does – it means voters are more likely to actually change their vote based on the issue, which means politicians are incentivized to focus more on what voters want on the issue (rather than focusing on other things, like what donors want).

And we’ve already established that voters want AI regulation.

5. Public First (likely) has a narrower set of priorities than LTF

Public First’s press release about their launch says they will “encourage AI transparency and safeguards”, with Chris Stewart emphasizing “meaningful oversight of the most powerful technology ever created” and Brad Carson emphasizing “guarding against risks to our children, jobs, national security, and humanity”. I expect LTF to be opposed to policies that do any of that.

But I also expect LTF to have strong preferences on policies that Public First won’t focus on. LTF will likely wind up in fights with datacenter NIMBYs and push for expanded energy production; I don’t expect Public First will oppose LTF there. LTF may also push for all sorts of policies that are unrelated to harms from AI but nonetheless increase the profitability of companies like OpenAI and Andreessen Horowitz. Again, I’d basically expect Public First to sit those battles out.

That is to say – LTF may have its firepower diluted by engaging in more issues than Public First.

6. Politicians always kinda knew LTF was a possibility, but Public First may be more surprising

Even before LTF was launched, politicians knew that major AI companies could spend tons of money on electioneering. Their political calculus would have already incorporated the implied threat of the possibility of something like LTF.

But the existence of a serious counterweight in Public First is genuinely more surprising. If Public First spends $50M to LTF’s $100M, I expect the main update for politicians will not be “wow, the AI industry outspent those who want to rein in the industry by a large absolute amount” but instead “wow, the pro-governance side of the AI policy debate raised a surprisingly large fraction of what the AI industry itself spent”.

Where from here?

Public First and its donors may now have an interesting choice. Assuming that Public First gets more bang for its buck than LTF, Public First’s donors could donate a fraction of what LTF’s donors give, and go toe-to-toe in terms of influence.

But they could also do something different.

Already in the NYT scoop about Public First, donors teased the possibility of matching LTF’s donors dollar per dollar. Given that potential donors for Public First appear to include various individuals with early investments in companies like Anthropic and OpenAI, as well as perhaps Anthropic’s corporate account itself, matching the dollars that LTF puts up should be doable. And given the fact that LTF was created first and is pushing policies against the public’s preferences, Public First would likely have a social mandate to match LTF in spending as a counter. So perhaps the best move for Public First wouldn’t be to match LTF in influence by spending a fraction of what they spend, but instead to match them in terms of dollars and then blow LTF out of the water in terms of effectiveness.