In early November, in that exceedingly brief window between the midterm election and presidential-primary season, some of the biggest names in the Democratic Party are expected to gather at the Four Seasons in San Francisco for a closed-door conference on artificial intelligence. There, a number of presidential hopefuls will sit for a series of talks and dinners with leaders of the country’s largest labor unions, donors to the biggest Democratic super-PACs, and political strategists who work with the party’s top elected officials.
The host of the conference is not a former president or party leader or name-brand interest group; it is Alex Bores, a two-term state lawmaker from the Upper East Side who lost the Democratic primary in June to succeed Jerry Nadler in Congress. Last year, Bores was mostly unknown in national politics. After sponsoring New York’s RAISE Act in the State Assembly — which, as of January 1, will require AI companies to publish safety protocols and report incidents of potential catastrophic risk in their models — he campaigned on bringing similar regulation to the federal level. This prompted Leading the Future, a super-PAC that has received millions from an OpenAI executive, to spend roughly $8 million on an effort to defeat him.
In This Issue
The Future of AI Will Be Shaped by a Few Dozen People
The donation cast Bores as a martyr to the political power of tech companies. There were profiles in Politico, Time, and Vanity Fair and a 90-minute sit-down on Ezra Klein’s podcast. Bores received a surge of campaign contributions as a rival set of AI companies and tech-world figures who support regulation poured $18 million into the race to counter LTF.
He ended up losing the primary to Micah Lasher, but in the process, Bores became perhaps the most preeminent voice for AI safety. It may seem like an odd fit: a politician with fame as an AI fighter who has major funders with ties to the industry. But the combination has helped Bores establish his next move. He is now launching an audacious campaign to push the field of 2028 Democratic presidential candidates to take on a new breed of corporate oligarch and write the rules for the most pivotal technology of the 21st century. “What’s the David Foster Wallace essay about how a fish doesn’t know what water is? We are going to be shaping the environment that these candidates are running in,” Bores says. “We can’t match Leading the Future tit-for-tat in spending, but that’s all they have to offer. We will have public opinion and public support, and we will be shaping the terms of the debate.”
Bores’s we is Who Decides, an advocacy organization he started just months after losing his race, when he and his supporters were having conversations about how to maintain the momentum from his congressional campaign. (Who Decides raised $10 million at launch and has planned commitments of another $20 million.) Part of the work will be traditional advocacy. Bores has already met with many would-be 2028 contenders — among them Kamala Harris and Pete Buttigieg — or their advisers. But he also has a bigger idea in mind.
Who Decides is hiring dozens of staffers and placing them in the six early-primary states and the seven key swing states to meet with “grasstops” leaders — prominent local pastors, Democratic Party county chairs, heads of Indivisible groups. By organizing among powerful on-the-ground constituencies, Bores thinks Democratic candidates in 2028 will have to go on the record with their positions on AI policies to gain support from the party’s base. “I want to make AI safety not like Medicare for All or ‘Abolish ICE,’ where half the people running are for it and half the people running are against it,” he tells me when we meet in early October. “I want to make it like marriage equality or being pro-choice, where if you’re a Democrat, of course you believe in AI safety.”
Bores explains this while ordering a drink at an Irish bar on Murray Street. He’d just come in from a nine-plus-hour City Council hearing across the street that was headlined by Jacob Coxon, the former Anthropic employee who announced on social media that the technology his company was building could destroy civilization. After the meeting, a number of activists, operatives, and politicians had gathered to drink beer and talk about a “national Democratic AI agenda for 2028.” The event, sponsored by Guardrails Action — another group working to push AI safety — was one of a series Bores has attended over the past few months.
As the Democratic presidential-primary race gets underway, party strategists increasingly view AI as the central tension of the contest, partly because AI touches on every other issue — wealth inequality, the economy, the environment, education. Concerns about data centers alone have morphed from a local land-use issue into a stand-in for the runaway power of tech companies and the possibility of an economy that makes most white-collar employment obsolete. And in part this is because it’s impossible to talk about AI without acknowledging that Coxon’s warning may prove prescient.
“Imagine the presidential campaign of 1944, and it’s 1942, and instead of Los Alamos, you have Ford, GM, Standard Oil all fighting to create not a single bomb to drop on Japan but an entire nuclear arsenal. And imagine that in 1940, Charles Lindbergh had been elected. And imagine that you have Hitler and Stalin also racing the three biggest companies on earth to build their own nuclear bomb,” says Daniel Squadron, co-founder of the States Project, which is dedicated to electing Democrats to state legislatures. “There is zero chance that ends well. And that’s where we are.”
So Bores finds himself tasked with getting Democrats to take on AI in a way that can appease both angry voters in the heartland and his big-tech donors. “We need to get to a place of ‘yes, and’ on these things,” he tells me once his beer arrives. “It is not ‘Oh, here’s the hymnbook.’ It is ‘Let’s all sit around a table and design the hymnbook.’”
When Bores ran for Congress, he liked to point out that if he won, he would be one of the only members of the House with a computer-science degree. He has become known for talking about the possible harms of AI, but he often discusses the benefits, too. “I want AI that helps doctors diagnose disease instead of AI that allows health-insurance companies to deny claims,” he told me when running in the primary. “This is a really potentially powerful tool, and we are on the cusp of having access to knowledge and power that we never had before.”
As more voters have become aware of the threats of AI, some in the tech world worry that the views of the far-left flank of the Democratic Party will become its de facto position. Bernie Sanders has called for a moratorium on data centers and further AI research. Others on the left think AI companies should be treated as public utilities. “We have to use this moment where there is fury at this idea that a handful of billionaire tech bros — who aren’t even offering us a product anyone really wants — might destroy the whole planet for their own trillions,” Brad Lander, the progressive former city comptroller who ousted incumbent Dan Goldman in the June Democratic primary for Congress, tells the crowd at the Guardrails Action event. “There is fury, and that is useful.”
Bores seems less keen to claim that no one wants the technology. His position is based more on oversight: requiring independent auditors, transparency on safety incidents, a ban on recursive self-improvement. Despite Bores’s elevation as a go-to speaker about AI’s worst outcomes, some Democrats think only someone who is explicitly anti-tech will suffice. “I don’t care who you are: If you say something positive about data centers, it’s going to be like saying on-camera that you wanted to abolish the police,” says Morris Katz, the Democratic operative. “You have simply misread the moment, and you are never going to be president because of it.”
David Shor, the head of data science at Blue Rose Research and something of a data guru to Democrats, has urged his party to notice that the national sentiment is not caution but rage. “People hate corporations. They hate the government, and they are extremely skeptical both of relying on the goodwill of these companies and of big government programs,” he said at a recent conference, citing research that shows broad majorities — even among conservatives — favor heavily taxing AI companies to help people who have been negatively impacted by the technology. “Voters buy into this idea that if you lose your job because of AI, that is the government’s problem and it is the government’s role to find you a new job and to make sure that you can continue to pay your bills and not lose your home.”
Bores’s former backers are less bombastic in both tone and approach. “You can’t go like Sanders,” Chris Larsen, a crypto billionaire who, through a PAC, gave Bores $3.5 million for his congressional run and has emerged as one of the leading funders of the party, tells me. “That goes too far. It’s just too simplistic, and it doesn’t really have a plan. It’s just sort of, Oh, we can just put AI in a box and drop it in the ocean and it’s all over. You can’t do that.”
This tension has made many Democrats cautious. Some worry about looking foolish by being too anti-AI. The technology is changing rapidly; the politics could too. What happens if the AI stock bubble deflates? What happens if China races ahead? Then again, what happens if Democrats don’t act fast enough? Liberals could lose the advantage on the issue after Donald Trump, who has catered to the AI industry, leaves office. Republicans like Steve Bannon and Missouri senator Josh Hawley have taken a much more aggressive approach to curbing the tech companies’ power.
“What concerns me is that people really fucking hate this issue when you put it in front of them,” says Rob Flaherty, a digital director for the Biden White House and deputy campaign manager for Kamala Harris in 2024, who has started his own super-PAC to support pro-AI-safety candidates. “And if we are weak and mealymouthed and then the Republicans do their demagogue thing — which would be about ‘tech elites’ and ‘Big tech is coming to fuck up your community’ — then that opens up a space for a conservative who is an economic populist.”
Bores has promoted Who Decides as the solution: an outlet for voters’ anger that is sensible enough to convince those who think like Larsen. But for his critics — especially supporters of Leading the Future — his attempt to find a middle ground only proves their point from the primary: Bores is not a legitimate voice for safety but a prop for one side of the debate. “He just got a $30 million bag from the same doomer groups who funded his campaign,” Josh Vlasto, a co-leader of LTF, tells me of Who Decides.
A longtime New York political operative who came up in Chuck Schumer’s office, Vlasto served as a top aide to former governor Andrew Cuomo. From there, he worked as a spokesman for Fairshake, the leading pro-crypto PAC. Vlasto maintains that whatever Bores may claim about taking on the power of the tech companies, his congressional campaign was supported by a different set of tech executives who warn of the perils of AI and call for strict controls because whipping up hysteria about the potential for doom is good for their share price.
“They are funding a narrative that is driving a threatening atmosphere to people who want to be thoughtful on these issues,” he says. “And we want to stand up for people who want to pass a smart bill and smart policies in the face of relentless pressure.”
Vlasto tells me this in his office in midtown as the dregs of a weekend nor’easter lash the windowpanes. When I arrived, he was barking into the phone, speaking with a client who runs a company that scans social-media to determine how much of any online conversation is coming from bots. A useful tool, he says, “when you are getting fucked with on social media.”
When LTF launched, it was labeled a Republican operation masquerading as a Democratic one. And for good reason: The group’s biggest funders are Andreessen Horowitz, the venture-capital firm that has given tens of millions of dollars to Republicans in the past few election cycles, and Greg Brockman, the president of OpenAI, who gave $25 million to President Trump’s super-PAC last year. Vlasto’s partner Zac Moffatt is a former Mitt Romney aide. LTF got involved in about as many Republican primaries this year as Democratic ones.
Despite this baggage, Vlasto says the notion that Americans oppose AI is a canard propagated by Bores’s side and a mistake for Democrats. He cites polling data that shows less than one percent of voters mention AI and data centers as their No. 1 concern this midterm election cycle. While loud calls for banning data centers get the most play on social media, he says most Democratic politicians want data centers to go forward as long as meaningful guardrails are attached. So far this year, Vlasto says, almost no Democrats are running ads about banning data centers. If the Democrats go anti-AI, or anti–data center, he says, they risk alienating onetime parts of their base, such as blue-collar building-trade workers. “You go to Nevada, you go to South Carolina, and run a campaign that says ‘AI is going to destroy the world,’ and all these data centers that have created union jobs in working-class communities — I mean, I’m just not sure these guys have a feel for what the politics in this really are.”
LTF’s counterpart during primary season was the super-PAC network associated with Public First Action. The group, the biggest supporter of Bores’s congressional campaign, was given $40 million by executives and engineers at Anthropic. Brad Carson, a former Democratic congressman for Oklahoma who serves as the super-PAC’s leader, disagrees with Vlasto that his donors just want to write rules that favor their own employer. “Say whatever you want about Anthropic, but they’re very sincere about wanting more regulation. They’re not like making it up or pettifogging,” he says. “I get why Josh is always saying this is an Anthropic-vs.-OpenAI fight. But if you’re starting a super-PAC and someone calls you up and says, for whatever reason, ‘We’ve decided we want to give $20 million to your cause,’ you say, ‘Okay, you have no impact on our governance. Sure, give us money.’”
LTF and Public First battled in congressional races elsewhere in the country in 2026, but the Bores race, held in the media capital of the world, got the most attention. Vlasto’s playbook called for running a campaign that touched on all sorts of issues except for the one he was actually being paid to push. It backfired in Bores’s race because of how many well-informed voters are in the district. “Manhattan is ground zero for global media, and when we started this campaign, AI wasn’t even on people’s radar,” Carson says. “But all the media people live in the district, and suddenly Ezra Klein’s covering it, Michelle Goldberg’s covering it, and it makes AI issues front and center for the American elite.”
Vlasto says Bores became LTF’s first target because, after Trump’s reelection, Anthropic was no longer Washington’s preferred AI company, so the firm turned to the states to regulate. In Vlasto’s view, Bores is no plucky upstart; he is a node in Silicon Valley’s vast network of effective altruists. “I am not naïve. I think these fucking EA guys are going to put a billion dollars in a pot, and Alex Bores and like-minded groups are going to go up to each of the candidates and say, ‘I have a billion dollars. This is the policy process we want you to follow,’” Vlasto says. “But I don’t necessarily think it is going to work.”
For LTF, beating Bores wasn’t about just one race; it was about defeating a “doomer” worldview that Vlasto says wants to slow the pace of progress — and a rival company that would benefit from any kind of slowdown. Bores disagrees. “Their goal was not just to beat me but to crush me and to make it so that no one else wanted to run on AI,” he says. “They failed horribly in both of their missions, and their donors have realized that.”
In a company Slack message in June, Brockman, the OpenAI founder who gave LTF $25 million, part of a $50 million pledge to the group, announced he would withhold the second half of his donation, acquiescing to pressure from OpenAI engineers who disagreed with LTF’s approach. Last month, a spokesman for Ron Conway, who was LTF’s leading Democratic donor, announced he too would no longer fund the group. For Bores and Carson and the rest of their side, this was a cautious victory.
“I am sure these guys are going to reboot, and it’s why you are seeing the pro-AI-safety wing get their shit in order,” Jesse Ferguson, a longtime Democratic operative who advised Bores’s congressional campaign, tells me. “These are the insanely super-wealthiest people in world history, and they are going to do everything they can to stop any regulation. And so we can defeat the super-PAC, but that doesn’t mean we have defeated the money behind it — their first line of defense is going to be the Republicans, and their second is going to be the hearts and minds and bank accounts of the Democrats running for president in 2028.”
Vlasto insists LTF is still moving forward. He says his group currently has $40 million in the bank. “This organization is funded as a multi-cycle organization, and I think we’re just going to keep going,” he says.
How this will play out in the 2028 primary remains uncertain. “It’s a game of chicken,” said one political operative who has been in communication with several would-be contenders. “No one wants to jump out ahead and be the AI-safety candidate. They know there is all this money up for grabs so they are all lying low and trying to be somewhere in the middle — not the best candidate for Silicon Valley but not the worst, either — and waiting to see where everyone else puts their stake in the ground. Leading the Future might be dead, but their money is very much alive.”
For now, no Democrat serious about running for president is lining up as an outspoken defender of the AI industry. But Bores still says he is on the lookout for any indication that a candidate has gotten too close. “I will start knowing a lot more when their plans come out, but if there aren’t hard limits in your plan, or if you are saying ‘We’ll keep talking about it,’ or if you’re doing dog whistles about how we have to beat China more than talk about how we have to protect Americans, and if you are giving this overly rosy story about how it will all work and only alluding to potential harms,” he says, “that’s a pretty good sign you are about to get laughed out of the primary.”
And he is confident that his plan for how to shape a candidate’s AI message will work. “They get their power from spending insane amounts of money and keeping advocates divided so they can pick people off bit by bit,” Bores says. “But if all the voters are saying the same things in every town hall, and if every group that you call is saying it supports regulation, then the cost for buying candidates off is gonna go way, way up. And that is what we are making clear.”