How Big Tech Lost The Midterms

For years, the policy debate over artificial intelligence belonged to those who counted progress in decades. An eccentric group of developers and small investors argued the technology’s potential payoffs — from booming productivity to solving death — were years off. And then, all of a sudden, their science fiction-inflected fantasies were real.

As AI capabilities vastly expanded, a cleave down the middle of the industry formed. Some in San Francisco and Washington latched on to the idea that regulating AI would risk a brighter future for humanity — or hand China the keys to global prosperity. Their opponents, many with a worldview shaped by the California-based effective altruism movement’s focus on risk to generations not yet born, felt more comfortable measuring their plans in centuries. They worried about the possibility that this nascent technology could lead to the end of the human race.

These philosophical quandaries that dominated cocktail (or mocktail) parties in San Francisco soon hurtled back to earth in Washington. In May of 2025, a budget reconciliation bill passed the House of Representatives with a provision that would have punished states for enforcing their own rules about AI models for a decade. The tech industry had grown deft at using its lobbying might in Washington to convince politicians to hold back from regulating its products. Now, with outsized influence within the Trump administration, some of the world’s wealthiest companies were emboldened to go further, promoting a so-called preemption provision to prevent states from doing any policy work on AI at all.

The provision did not become law — it was voted down 99-1 in the Senate on the first day of July — but the month-long fight over the measure permanently changed the way those on all sides of the AI policy debate approached their work. Many who had believed they were operating on a timeline of decades and centuries were suddenly focused on a much shorter horizon: the 16 months before Americans would determine control of Congress at the polls in November 2026.

The more anti-regulation side of the AI industry — represented at the time by venture-capital firm Andreessen Horowitz and OpenAI — was already working to establish its first super PAC, explicitly modeled after a pro-cryptocurrency effort that had spent big in a few races to instill fear in lawmakers who had crossed them. Leading the Future, as the new AI committee was named in August, would commit itself to confronting what its leaders called “a vast force out there that’s looking to slow down AI deployment.”

Their decision to try to thwart any AI regulation and then brashly enter the world of electoral politics would have an unintended consequence: motivating that vast but scattered force — a mix of tech policy veterans, philosophically minded effective altruists, consumer protection advocates, anti-monopoly groups and religious right organizations — to work together for the first time. An impromptu alliance that had rallied to defeat the preemption provision joined forces as they, too, looked for the first time to the electoral calendar.

“It congealed over that fight, and it has sustained since then,” said Brad Carson, a former Democratic congressman who leads Americans for Responsible Innovation.

Carson, a 59-year old former Rhodes Scholar who represented Oklahoma in Congress before being awarded a Bronze Star for his service in the Iraq War, began plotting the creation of a group that could spend big money to counter LTF. Jay Shooster, a proudly nerdy 36-year-old consumer rights attorney, looked to fund candidates who brought existing credentials around AI policy. Rishi Bharwani, a 30-something Democratic operative who teaches yoga when not working on campaigns, saw a chance to change the way Democratic candidates just discovering the salience of the issue talked about it.

“I had no idea how quickly we would get to where we are now,” said Michigan state senator Mallory McMorrow. In 2025, according to McMorrow, the legislature had “barely scratched the surface” of any sort of AI regulation. But just a few months later, with Shooster’s guidance and Bharwani’s messaging, she came out with one of the most detailed plans on AI safety in the country as a U.S. senate candidate.

Carson, Shooster and Bharwani have found themselves thrown together in a coalition that has proven itself to be remarkably effective this cycle, especially on the left. They have shepherded a tidal wave of pro-AI safety cash that flooded the zone across the country, picked out targets, expanded the map on AI issues and scored major victories.

And yet, they represent three very different interests. Carson has spent his time trying to beat back AI industry cash — while getting donations from another part of the AI industry. Shooster is worried primarily about existential risk from AI, building a network of small-dollar donors to make an impact on the left. And Bharwani, who is far less involved with AI safety in Silicon Valley, has spent his energy trying to convince more Democratic politicians to take a hard line on child safety online.

Their divergent approaches to the midterms are indicative of larger philosophical differences in how to fight and win on AI policy and politics. Their coalition is already cracking, and may well break apart in the fog of the 2028 primaries. But first, it delivered one of the quickest and most stunning reversals in fortune in American political history.

“I’ve done work over the years on prescription drug pricing,” said Jesse Ferguson, a Democratic campaign strategist who served as a senior adviser to one of the cycle’s first congressional candidates to make AI safety an issue. “Twenty years of people in both parties being afraid of pharma and the drug companies. This time around, that fear went away in six months.”

Carson was sitting in a Tulsa lesbian bar called Yellow Brick Road in August 2025 when his phone lit up with the news that the pro-tech faction had launched LTF, their own super PAC.

Years earlier, Carson had become interested in artificial intelligence after attending a conference hosted by the Future of Life Institute in San Juan, Puerto Rico that included some of the now-biggest names in AI, including Anthropic’s Dario Amodei and OpenAI’s Sam Altman. While still serving as the president of the University of Tulsa, Carson launched the AI nonprofit Americans for Responsible Innovation.

By 2025, he was well connected in AI safety circles — and quit his plum post at the private research university to focus on AI issues. At ARI, Carson focused on trying to shepherd AI safety regulations through a hostile Congress. The preemption fight was his first major undertaking, and he helped to stitch together over 100 disparate organizations to fight together. He namechecked Bharwani and Reset Tech as another one of the major contributors to the effort.

Seven weeks later, Leading the Future launched. The organization came out filled with swagger, confident in its ability to leave an imprimatur on the midterms and beyond. With $100 million in the bank and plenty of donors to tap for more, combined with the lack of care about the issue from the American public, the AI industry looked like it would be able to bankroll its preferred candidates to victory.

“[We’re] a counterforce to … existing spending to push one company and one ecosystem’s values and corporate agenda,” Josh Vlasto, a co-leader of the group, told POLITICO Magazine in July, referring to Anthropic and the company’s friends in San Francisco. “We view them as trying to derail a thoughtful policy process.”

When Carson’s phone buzzed with the news, he got nervous. He had met plenty of rich people who wanted AI regulation. But they were more accustomed to propping up networks of nonprofits than getting involved in electoral politics — which many in Silicon Valley considered a dirty game filled only with trouble. That view was only reinforced after Sam Bankman-Fried poured cash into races across the country before his downfall.

In response to the news large networks of people working on AI safety started texting each other. They had come to a similar conclusion as Carson had — that at best, this was an ominous development.

The world of AI safety does much of its thinking and operating via group chats on the messaging app Signal. The group of people spending their time pushing for AI regulation pre-2025 was relatively small — and often remarkably wealthy. Carson began to tap those connections to create Public First, a super PAC network that would be a clearinghouse for pro-regulation cash.

In the early stages of Public First, Carson was not working directly with the AI company Anthropic, he said. But he was coming down on one side of a fight between Anthropic, which was more pro-regulation, and OpenAI, which had a more accelerationist posture. The two AI companies, in addition to disagreeing somewhat on regulation, are also locked in a broader competition, with real animus between Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman.

“The public is on our side about this,” Amodei said at a September 2025 kickoff at Washington’s Union Station in which Anthropic formally announced its decision to get involved in AI regulation lobbying efforts. “We don’t stand alone at all. It’s the tech companies that stand alone, no matter how many of them there are.”

For Public First, their original goal was to see if they could raise $5 million from those AI safety advocates, Carson said. Then it became $10 million. Then $30 million.

The people donating to these efforts — from Public First and otherwise — have a similar worldview but often very different job titles.

“It’s a very disparate network of people in the safety space that includes people who have started companies,” said a Democratic political strategist who advised multiple Senate and House campaigns this cycle, granted anonymity to speak candidly about donor dynamics. “But [it] extends down to a lot of random Anthropic engineers who got hired there in 2022 and are now centimillionaires and want to give in this space, but have zero interest in any kind of public persona.”

In February, Anthropic decided to get involved with a $20 million donation from its corporate account to Public First’s associated nonprofit — which can’t directly back candidates or engage in electoral politics. In May, Amodei gave his first-ever public seven-figure political donation — of exactly $1 million — to Public First. And in July, Anthropic donated another $20 million. That same month, Carson says the AI company had considered starting its own safety nonprofit and PAC. Once Public First came on their radar, they decided to direct their money to an ally instead.

Anthropic declined to comment.

Before Public First, before Leading the Future and before preemption, there were very few politicians around the country who wanted to talk about AI at all. One exception was Shooster, who ran for the Florida statehouse in 2024 in Boca Raton.

In his own race, Shooster would talk to Jewish grandmothers in Boca about the potential downsides of AI growth — especially after his own family almost fell prey to an AI scam. But more important was his base of donors: EAs and AI engineers with deep pockets.

In November 2024, Shooster watched in dismay as Donald Trump swept back into power. He wanted to take a break from politics. But friends working in AI safety policy convinced Shooster that “AI capabilities were about to get scary, and it was a now-or-never moment to do something about it.” Soon after, he moved from Florida to San Francisco. And along with his friend Eric Neyman, Shooster started to bundle hard-dollar donations from those who donated to his own campaign, along with a much broader network of people in Silicon Valley worried about risks from AI development.

In January of 2025, Shooster trained his attention on one candidate: Democrat Manny Rutinel, who announced he was running almost as soon as Trump arrived in office. Rutinel, who co-sponsored an AI consumer protection bill in the Colorado Senate in 2024, has been described as “the politician with the closest EA ties of any member of Congress, ever.”

That same month, Shooster announced that he would serve as the chair of Rutinel’s finance committee and called him a “dear friend.”

For many AI safety advocates, having people in Congress who already understood the issue of AI regulation was essential.

Shooster sought to help Rutinel in part via bundling, the practice of soliciting checks to the campaign from individual donors and presenting them together as demonstration of coordinated support. The amounts would be relatively small compared to the eight-figure donations flowing to Leading the Future, which as a super PAC could accept unlimited sums from individuals as well as direct gifts from corporations.

And yet, there was power in individual contributions — especially within a closely aligned network of wealthy people in consistent, close contact.

Shooster and Neyman knew their asset was a network, cultivated over years of coalition-building for other advocacy work. Soon, it wasn’t just a bunch of tech workers who wanted in the fight. It was some of the biggest names in AI — who were rapidly getting richer thanks to the growth of the products they were so desperate to regulate.

“Convincing the first people to start giving was a bit slow,” said Shooster. “But then we reached a critical mass where political donations started to become seen as a normal giving opportunity, and it really took off.”

In the same week in October, two more Congressional candidates with major bona fides in AI regulation emerged: New York State Assemblymember Alex Bores and California State Senator Scott Wiener, who had both written state legislation concerning regulation of AI models and had been enmeshed in fights with Democratic governors to get those bills through. Both were running in deep blue, very wealthy districts with a high concentration of media professionals — and running against someone with deeper ties to the more traditional Democratic machine.

Shooster watched as Leading the Future quickly staked its reputation on keeping Bores out of Congress, eventually spending over $8 million against him. One person with direct knowledge said that LTF co-head Josh Vlasto told prominent Democratic donors that “we will spend whatever it takes to keep Alex Bores out of Congress.” They decided not to engage with Rutinel or Wiener.

“It’s an uncanny, weird coincidence that the two biggest legislative champions for AI safety in the entire country announced their bids for Congress just two days apart,” Neyman wrote in a remarkable 8,093 word post on an EA-friendly blogging site called LessWrong. (Another post asking followers to donate to Bores two days prior came in even longer, at 8,494 words.)The money started to flow, both into AI safety super PACs like Public First and directly into the campaign coffers of Bores and Wiener. Bores reported raising over $1.2 million on his first day in the race; much of that was from the bundling efforts of Shooster and Neyman. (Bores’s leading primary opponent, state assemblymember Micah Lasher, was a co-sponsor of Bores’ AI regulation bill, but was not connected to the AI safety world.) Leading the Future had hoped to make an example of Bores.

A race that included a Kennedy scion and a famous anti-Trump commentator seeking to replace a quarter-century incumbent and former Judiciary Committee chair suddenly became a proxy fight over AI policy. Single-issue donors transformed the race ahead of the June 23 primary, with pro-regulation political groups ultimately spending over $19 million in favor of Bores — swamping LTF’s $8 million. Over $13 million of that pro-Bores money came from Jobs and Democracy PAC, an organization in the Public First umbrella.

A deluge of pro-Bores mailers and TV spots blanketed the city and the airwaves. It was impossible to watch a Knicks playoff game without seeing (extremely expensive) ads about Bores.

“We showed that there is political upside to taking this position, and that includes financial, though that was not at all obvious at the start of that race,” said Bores. “The vast majority of that [pro-regulation] spending came in at the very end.”

Bharwani, who was not deeply involved in the Bores primary, was baffled at how it became such a flashpoint and at why Leading the Future spent so much money on the race.

“It was a crowded congressional primary in deep blue, upper middle class New York — not really symbolic for the country or representative of a swing state. But [LTF] chose a person they wanted to take out and claim credit for,” he said. “Why they chose Alex Bores, a thoughtful, charismatic tech policy expert who is very handsome, is beyond me.”

His comments speak to how his thinking about AI and tech regulation diverges from both LTF and from Shooster and Carson. Leading the Future saw Bores as a shill for pro-Anthropic interests. Silicon Valley advocates saw him as essential to advancing said regulations in Congress, given his background and understanding of the issue.

Bharwani, meanwhile, has focused his efforts on shaping the views of candidates without a background in AI policy. He thinks he can get them to come around to his wants.

As the New York primary entered its final weeks, Shooster and Neyman welcomed another promising Democratic candidate to California. Like Bores and Weiner, McMorrow was an energetic young state lawmaker challenging a favorite of the local Democratic establishment in a race for federal office. Unlike them, she had a scant record engaging with policy questions around AI.

McMorrow had voted for a bill that required political advertisements to disclose AI use in the Michigan Senate in 2023. But she had little experience with AI development, and her nascent campaign for U.S. Senate was focused on other issues — high prices and fighting corruption in the Trump administration.

“We had barely scratched the surface [of AI] in the legislature,” said McMorrow.

But as she traveled around Michigan, McMorrow grew more curious about the technology. After a voter at a trade union breakfast talked to her about his fears of AI surveillance, she started to think about whether the issue was becoming increasingly important to Michiganders — and whether she could run on more than just chatbot safety and regulations for data centers. She and her staff began to see regulating AI not only as responsible policymaking, but also as a political opportunity: a way to distance herself from her primary opponents.

A few months later, she traveled to San Francisco to raise money and meet with multiple groups trying to prevent catastrophic risks to humanity from AI. Many of those meetings were organized by Shooster and Neyman, according to multiple people with direct knowledge. (“We’ve been grateful to host a number of candidates and elected officials for conversations with AI safety experts, and look forward to hosting even more in 2028,” said Shooster.)

At the same time, other ambitious Democrats were also growing queasy about tech investment they had once invited. In February, Bharwani received a call from Pennsylvania Governor Josh Shapiro’s office. A year earlier, the Democrat had welcomed “state-of-the-art data center campuses.” But as he entered a reelection year, he began to rethink his open-arms policy to Big Tech investment. Seeking a more expansive way to address AI risks, Shapiro advisers wanted Bharwani’s advice on how to focus on chatbot harms to children and reached out as part of broader discussions with stakeholders. The same month their offices spoke, Shapiro highlighted child safety concerns in his annual budget address. A few months later, the Pennsylvania governor sued Character AI, a generative chatbot and AI roleplay platform, over unlawfully presenting itself as a medical professional. The company recently declared bankruptcy.

Shapiro managed to quickly pivot. But people like Shooster, who fear AI could cause cataclysmic events, feel that politicians should be talking about more than just data center construction and chatbots interacting with children.

McMorrow’s vision of a comprehensive approach to AI policymaking — one concerned with surveillance and the physical presence of data centers as much as guardrails on model development — was met with skepticism from people working on and researching these products. They were focused almost entirely on one thing: They believed that their work, if not regulated properly, could lead to the destruction of humanity. If taking a position on data centers was the moderate thing all Democrats felt comfortable with, talking about how AI could kill all of us wasn’t so easy.

McMorrow said she “went down the rabbit hole” with them. But she also sounded a few notes of caution. “In my role, there is a danger that people will crater if [the problem] is too big,” she continued. “You look at it and get overwhelmed, and say, ‘If this is going to kill us all, then what do I do?’”

She saw the data-center issue as a way of making the abstract concern over catastrophic risk tangible for people today. Loud buildings landing uninvited near their homes could lead to voters to bigger conversations about what AI could do to the world.

“I remember some feedback in the room that ‘this isn’t about data centers. People who are making it about data centers are missing the point,’” McMorrow said. “I countered that, for the average person that I’m dealing with every day, their first entry point was data centers. If we aren’t able to connect the two things, you lose a lot of people.”

Bharwani advised her campaign on how to merge Silicon Valley advocates’ worries about annihilation with planks related to worker retraining, data center regulations and children’s safety. Back in Michigan, McMorrow developed multiple different plans that she released in separate documents under a broader “People Over AI” banner in early June. One initiative tackled AI safety and existential threats. A day later, with around 100 people packed into a small brewery in Elk Rapids, McMorrow talked in detail about her new initiative in response to a question from a Presbyterian minister worried about the technology, according to a reporter from AI-safety newsletter Transformer News who attended the event.

“For over a year, I stopped at breweries all across the state. People could ask me whatever they wanted,” said McMorrow. “The fact that this came up unprompted? Way up in very northern Michigan? From a pastor? That showed me this is going to be a big issue, and it’s going to cut across everyone.”

That same week, Wiener successfully advanced to the general election in California, winning a 40% plurality of voters and putting himself in the driver’s seat for a November clash against progressive Connie Chan.

Later in June, Carson, Bharwani and Shooster were all disappointed when Bores lost his race. But the loss appeared to presage a brighter future: A coalition focused on AI regulation had marshalled an almost unlimited amount of money to confront pro-AI cash. Despite the electoral defeat, it proved to Carson and Shooster that AI safety money could overwhelm the other side.

“We showed our might,” Carson said. “We outspent [LTF]. … It showed people that if you want to take this position, you’ll be defended.”

“For us, it was the end of the war.”

As the summer heated up, so too did the cash flowing towards AI safety groups eager to play a part in the midterms. The issue quickly became popular in progressive political circles.

New groups started to emerge, as well. In June, the progressive pro-regulation group Guardrails Action launched with $5 million raised. In September, Biden administration veteran Rob Flaherty and former Bernie Sanders aide Karthik Ganapathy announced Tomorrow Together, with over $10 million already raised. That same month, Shooster and Neyman officially launched an advocacy group called Citizens for Safe AI. Their website consists only of a landing page with the name of the organization set against a white background.

One of the season’s biggest announcements came from Bores, who spun up his own political advocacy group called Who Decides, along with his top legislative aide Anna Myers.

“We did fall short in that race, but we took the lessons of that coalition and fed it into what we’re doing now,” said Bores. “We saw candidates throughout the country take up the cause.”

The groups were likely drawing funding from a similar pool of AI safety advocates. Some, including Guardrails, were also trying to raise cash from labor unions and other progressive organizations. (Because of the way all of these committees were set up, it is impossible to know exactly who their donors are. Each has a 501(c)(4) nonprofit organization where donors can give money later funneled into super PACs that can spend directly on elections.)

They all benefited from a sudden late summer freakout over AI. In September, an unknown researcher named Jacob Coxon resigned from Anthropic; his announcement went mega-viral and led to a panic in Washington about the need to regulate AI systems. The news cycle eventually died down, but it left more attention on the industry in its wake — and a voting public more attuned to issues of AI and data centers.

“Jacob Coxon and the Hugging Face hack made the theoretical risks of AI real, alongside an administration and a president that was again and again taking the side of the deeply unpopular companies,” said Bharwani before the election. “Those factors together have allowed for this to be one of the biggest issues we’re seeing candidates take with them on the trail during the closing stretch.”

That attention helped another Michigander who had been carefully planning for an AI backlash for months. Even before he launched his bid against first-term Republican Rep. Tom Barrett in August 2025, Will Lawrence had been approached by a friend who needed help. A group of locals — mostly active on Facebook — in nearby Howell Township needed help organizing against the development of a data center.

Just a few weeks after launching his congressional bid, Lawrence attended a community meeting in the local high school’s auditorium about a local data center site. Town officials had already moved the meeting from a smaller venue to accommodate what they expected to be a large crowd. Still, the auditorium was packed.

As the campaign season ramped up, so did proposals from companies to build data centers across his central Michigan district — in Mason, Lyon and Lansing. In June, as Lawrence’s campaign planned its first campaign ad, the message clicked: It would be focused almost entirely on data centers.

In the ad, Lawrence, wearing a loose collared shirt and zip-up hoodie, stands in front of an empty field that he says is the site of a potential data center. As a shadow expands across the greenery, the Big Tech threat becomes figuratively front and center.

“This is really about what it means to be human, and what is the role of screens and what is the role of AI and of the online space in our lives,” said Lawrence, who founded the progressive climate group Sunrise Movement. “It’s a marvelous expression of human ingenuity and technology, but it poses profound questions about what it means to be human, and we risk becoming utterly dependent on this technology in a way that is hard to reverse.”

His campaign impressed McMorrow.

“This is an issue where Will Lawrence can make a very, very clear case against Barrett, and it resonates with people,” she said. “Particularly in Mid-Michigan, where I know there are a number of projects that are being considered that people are furious about, especially because of the lack of transparency and open meetings.”

Across the country, this is the cycle where AI politics have come online. According to analysis from Bharwani’s Reset Tech, between 2020 and 2024, just six ads used the phrase “big tech” or otherwise attacked the tech industry. This cycle, it’s 387. Across the board, candidates have spent $103.6 million on those ads.

Between 2020 and 2024, zero Democratic ads mentioned data centers. In 2025 and 2026, it was 659.

On the first Monday in October, an actor named Pat Dwyer sat on the second floor of a dusty office building in midtown Manhattan, looking the part of a human resources middle manager on his fifth firing of the day.

“So listen, we’re going to have to let you go,” Dwyer said as he stared into a handheld camera. “You’ve done great here, but we think we can get it done cheaper with an AI.”

The footage never aired. But it was part of multiple ad campaigns shot by Tomorrow Together for an ad buy totaling $4 million in House districts of Republicans who voted in 2025 to preempt state regulation of AI.

That ad buy included a $1 million spend against Barrett, the Republican incumbent being challenged by Lawrence.

“We had a pretty tight criteria,” said Flaherty, the co-founder of Tomorrow Together. “It had to be a presidential battleground state, had to be talking about AI in some way, and had to be a Trump district [that he won in 2024].”

The strategy was developed in part by growing ranks of operatives trained to work specifically on AI safety issues in campaigns. Coefficient Giving, a nonprofit funded in large part by megadonors Dustin Moskovitz and Cari Tuna — themselves closely connected with the effective altruist movement — recently launched a program to train political operatives who want to work in AI safety.

“I started to understand AI as … seeping into everything else that matters to me,” said Maddie Kriger, who participated in that program and had previously worked on national Democratic campaigns and in climate politics before joining Tomorrow Together.

Communication between political professionals not always versed in the technical side of AI and AI safety advocates is not always simple.

“I think sometimes hard AI safety people don’t understand what we do, and we don’t understand AI. Effective altruists meeting with ineffective non-altruists is always kind of funny,” Flaherty said with a laugh.

But those effective altruists often have deep pockets. And throughout 2026, AI safety advocates won the public relations battle on the left and the money battle in the midterms. AI safety-affiliated groups spent $44.4 million, compared to $27.9 million from the pro-innovation side. (That calculation does not include political giving by Elon Musk, the owner of xAI, although his America PAC did not appear to be focused on AI-related issues, nor does it include spending from 501(c)(4)s that is ostensibly educational in nature. Both pro-safety and pro-innovation factions spent millions through those sources.)

In September, OpenAI co-founder Greg Brockman backed out of a second planned donation of $25 million to LTF. In October, Jay Carney — who leads communications for Democratic megadonor Ron Conway — said that his boss regretted his LTF contribution.

“Ron, if he had to do that over again, he wouldn’t make that donation [to Leading the Future],” Carney told POLITICO.

On Tuesday, those efforts came to fruition. AI safety’s biggest advocates, Rutinel and Wiener, won tough races. All four House districts in which Tomorrow Together spent flipped from Republican to Democrat — including a narrow victory for Will Lawrence in Michigan. They also spent in two Senate races — in Ohio and Kansas — where Democrats won. The Guardrails Alliance won all five of the races that it spent in.

“The message around AI proved very effective as a connective point between cost of living and corruption,” said Flaherty. “Absent any big change in the policy regime of the United States of America, I think people are just going to get angrier and angrier about [AI] as we move on.”

Leading the Future did notch victories, including defeating Public First-backed Republican Scott Singer, who was running against Democrat Jared Moskowitz in Florida’s 25th House district. They also lost a major race in South Carolina, where LTF-backed Republican Darline Graham fell to Democrat Annie Andrews, and in the Kansas governor’s race, where their preferred candidate Republican Ty Masterson appears set to lose a shocker.

“We’re proud that many of the pro-innovation champions we supported this cycle won on Tuesday,” said Jesse Hunt, a spokesperson for Leading the Future. “We are committed to a multi-cycle effort and 2028 will show who actually cares about passing a national regulatory framework on AI and who wants to politicize an issue that’s extremely consequential to the future of our country.”

AI-safety campaigners are still taking a victory lap. A technology that was in no way legible as a policy issue at the start of 2025 had become a central question in American politics. When NBC exit pollsters asked voters about AI and data centers, 56% of respondents said that they were a crucial or major factor in their vote. That ranks well behind an issue like government corruption, which 90% of voters called a crucial or major factor. But over half of voters claiming that AI and data centers are important to their vote suggests that the issue was an important factor in the midterms.

“Trump courted support from Big Tech CEOs, cheered on data center developers and gutted the federal enforcement agencies responsible for holding tech companies accountable,” said Bharwani. “That left a lane for anti-tech progressive populists that was wide enough for Democrats to drive a Mack truck through. And they did.”

That coalition is well-positioned for the 2028 cycle. Not only is advocating for AI regulation clearly good populist politics, it now comes with a built-in set of donors and super PACs — plus a progressive political class who see work on AI safety as both a moral imperative and a career advancement opportunity.

Yet as Democrats in Congress try to figure out how to govern — and begin what is sure to be a messy presidential primary — Carson, Shooster and Bharwani might find themselves once again with a set of divergent interests. “The actual challenge isn’t going to be ‘do we all agree,’ it’s going to be ‘what is our lane?’” said the Democratic political strategist.

Carson is connected to one of the world’s wealthiest and most powerful companies, one that will be subject to new pressure from investors when it delivers on a promised IPO in the next few weeks. Recently, he carefully distanced himself from draft legislation on AI safety that Anthropic was proposing, arguing “the bill text that’s circulating does not meet the minimum requirements necessary for federal guardrails to keep AI development safe.”

Shooster’s influence comes from the ranks of EA-influenced donors, many of whom stand to become even more wealthy when Anthropic goes public. “Our organization’s top priority is addressing catastrophic risks,” said Shooster, while noting that goes “hand-in-hand” with addressing the ways that AI is hurting people in the more immediate term. For Bharwani, the already existing harms from chatbots or data centers are most of what matters.

“In their first 100 days, they should ban surveillance pricing, crack down on online scams, and stop data centers from jacking up electric bills. And they should draw a sharp contrast with Trump on child safety,” he said. “Democrats should pass kids’ online safety legislation and dare Trump to veto it.”