New York Magazine: The Democrats’ Mad Scramble to Get AI Right
Just before his paternity leave last fall, Representative Greg Casar of Texas was given some light reading — less Goodnight Moon, more “good-bye, humanity.” “A young staffer handed me If Anybody Builds It, Everyone Dies,” says Casar, referring to a book co-written by researcher Nate Soares that argues against building artificial superintelligence. Mulling it over “between the joy and the sleeplessness and the diaper changing,” the 37-year-old new dad says he began to think more about “what kind of world my kid would grow up in.”
At the time, his fellow lawmakers weren’t doing much to shape the AI future; labs like OpenAI and Anthropic operate without federal legislation to stop them from nuking the labor market or maybe even killing us all. Casar says he realized “how little leadership there was within our own party and certainly in the Congress on AI.” He came back from leave on a mission to warn his colleagues and voters about the possibility of disaster. “This is probably the most important thing that I and many others could be working on in Congress right now,” he says.
Not everyone agreed. As the year wore on, lawmakers did little as the big labs raised the cost of everything and Anthropic boasted of an AI agent so powerful they couldn’t release it. Then came the Hugging Face incident. In late July, OpenAI disclosed that hundreds of its agents hacked out of a closed experiment and into a company’s servers, then coordinated with one another to cover up the trail. The next day, Casar ditched his prepared remarks during a meeting of House Democratic leaders and brought up the hack. “Most people hadn’t heard, and those who did still didn’t fully understand,” he says. “As I laid the details out, many of the Congressional leaders who paid some attention to AI policy said, ‘You know, I thought we were years away from anything like this happening.’”
It was the start of a race behind the scenes for Democrats to understand the rapid advancements in AI — kicked into high gear when the researcher Jacob Coxon resigned from Anthropic in early September. The 27-year-old offered a warning on X: “The people building AI earnestly believe that it could kill us all by the end of the decade.” With the president calling AI threats a “hoax” and initially dismissing the idea of industry guardrails, Democrats had an opening. But they had to catch up to the moment.
“As the rest of Congress is waking up, Casar has been well placed to really help educate a lot of the rest of the body,” says Alex Bores. The 35-year-old New York state assemblyman and former data scientist was hoping to be in a similar position come next year, educating his older lawmakers on AI. But he lost his near-record-expensive primary in Manhattan; a pro-AI PAC spent $8 million in attack ads on the candidate calling for strict regulation. Bores has been on calls with Casar recently and chatting with a handful of likely candidates in the 2028 presidential run who are in “listening-and-learning mode” right now. Casar is also in high demand among the 2028 Democrats. A source familiar with the conversations says he has been in touch with six of the household names considering a run.
The AI doomers are getting meetings too. Soares says he used to fight to get attention from lawmakers. But last Tuesday, he had a meeting in D.C. planned with one representative; it grew as more and more members stopped in. “We got a conference room and kept the door open and we must have had six members coming through frantically asking a bunch of questions,” Soares says — basics like “How much time do we have?” and “What do we do?” It seemed to him like it was the “first time that they were asking those questions.”
Representative Bill Foster, a high-energy physicist before getting elected to Congress in 2012, has concerns that some Democrats aren’t getting things on a technical level. He says a large number of his colleagues just have a “list of buzzwords that they can repeat and sound like experts.” But he appreciates those who are trying, like Representative Don Beyer, who he has seen carrying a Python textbook.
If some Democrats are failing to fully grasp the problem, Foster says it doesn’t compare to what he is seeing from the White House. He insists the technical expertise matters. “I don’t think there’s anyone in the White House that can read the data sheet and specifications for a GPU,” he says, referring to the chips that power AI. This comes into play when negotiating how many Nvidia chips should be sold to China. “It’s the same thing as doing nuclear negotiations with Iran with a bunch of real-estate developers,” Foster says.
Some lawmakers were already engaged on the issue a majority of Americans are freaked out about. Democratic representative Lori Trahan from Massachusetts, with her Republican colleague Jay Obernolte, proposed a bill in July, just before Hugging Face, to oversee “the development and deployment of frontier artificial intelligence.” California passed a law in early September establishing third-party evaluators to monitor labs like OpenAI and Anthropic. In New York, a law goes into effect in January forcing these labs to register with the state and file quarterly reports on the kind of “catastrophic risk” Soares warns of. On Thursday, Senator Mark Warner proposed a bill that would require mandatory testing to ensure the safety of new AI products prior to launch. Senator Bernie Sanders spoke last week at an event with Steve Bannon, unlikely headliners calling for AI controls. Casar was there, too, warning that beating China in the rush to build superintelligence was not worth potential catastrophe. “Far too many people seem to be trying to win a race where millions, if not billions, of lives could be lost at the finish line,” he said in his speech. “Our goal must be to end the AI arms race.” On Wednesday, Sanders and Casar announced a bill to ban superintelligence and create a Cabinet agency to regulate AI.
Foster sees the kind of outright bans Casar is calling for as counterproductive, though he did sign a letter from Casar to OpenAI demanding more information about the Hugging Face incident. If Democrats take the House in the midterms, expect a lot more letters — backed with subpoena power. Politico reports that Democrats, led by Foster and Representative Ted Lieu, are considering a select committee on AI that could subpoena witnesses like Coxon or the executives themselves.
Still, Rob Flaherty says Democrats have to move faster to catch up. “They’re getting smarter, but you’d be surprised at the level of important person who has very few sophisticated thoughts about this thing,” he says. In 2024, Flaherty was the deputy campaign manager for Joe Biden and Kamala Harris; for the next cycle, he is advising Democratic candidates on how to approach AI. There’s a lot of work to be done. “I have had elected officials say to me, ‘Maybe we need some plumbers. We should train more people to be plumbers,’” he says. “I also heard, ‘I don’t think AI is going to have negative impacts,’” he says. “That’s a person running for president who said that.”