As AI fears grow, lawmakers are racing to turn alarm into action
By Adele Rutherford · Reporting from Atlanta ·
The problem with treating an existential risk as a political football is that the ball eventually pops.
The problem with treating an existential risk as a political football is that the ball eventually pops. As Jim VandeHei recently laid out in a proposal reported by axios.com, we are facing a "generational challenge" in the AI race, and by every metric of statecraft, we are blowing it. VandeHei is calling for a special session of Congress, a coordinating working group of the "best and brightest," and a governing body with the actual authority to shut down models that exhibit dangerous behavior.
President Trump, predictably, opposes such regulation and the bipartisan cooperation required to build it. This is not a mere disagreement over the pace of innovation; it is a fundamental failure to recognize the nature of the tool. When you are dealing with a technology that can rewrite the social contract or facilitate bioterrorism, "deregulation" is not a strategy—it is a surrender to chaos. We are repeating the mistake of the 2008 Financial Crisis, where systemic risks were ignored by regulators until the entire house collapsed.
The only precedent for this scale of urgency is the Manhattan Project. That was not a "regulatory" effort in the sense of red tape; it was a state-led, multi-sectoral mobilization of scientific and industrial resources designed to secure a decisive technological advantage in an existential race. The Manhattan Project succeeded because the government didn't ask the market to self-regulate the atom; it seized the process, coordinated the experts, and managed the risk from the top down. To treat AI as a series of product launches rather than a Manhattan Project-level event is to mistake a tidal wave for a swimming pool.
The Labor of the Invisible
There is a specific kind of competence that happens "behind the curtain," far from the cameras and the campaign rallies. I am reminded of the reporting from iatse regarding the twenty-fifth anniversary of the September 11 attacks. In the wake of that horror, while the political machinery was still grinding, the technicians of Local 52 and the Teamsters were simply doing "what they knew how to do," providing the lighting and rigging that allowed rescue operations to continue twenty-four hours a day. They didn't wait for a bipartisan consensus to provide light to Ground Zero; they applied a professional standard to a crisis.
VandeHei’s proposal for a working group—incorporating labor, ethics, public health, and AI executives—is an attempt to institutionalize that kind of competence. He is arguing for a coordinating infrastructure that maps problems before they become catastrophes. We saw the utility of this model during the COVID-19 pandemic, where public-private partnerships bypassed traditional bureaucracy to develop safety standards at warp speed.
The opposing argument, usually voiced by the Silicon Valley set and echoed by the current administration, is that any governing body with "shutdown power" would stifle American innovation and hand the victory to China. They argue that the "race" requires total freedom of movement. But a race where the participants are blindfolded and running toward a cliff is not a race; it is a mass casualty event. The strongest version of the pro-innovation case fails because it assumes that "innovation" is a vacuum. In reality, the most sustainable advantage is not the fastest model, but the most stable one. A US-China framework for AI control, modeled after nuclear weapons governance, would not hinder the race; it would ensure there is a finish line left to cross.
The Cost of Apathetic Truths
The friction currently residing in the White House is telling. Vice President Vance is described as having "rising worry" regarding AI, yet he remains tethered to a President who views regulation as an inherent evil. This tension is a microcosm of a larger American malaise: the tendency to withhold necessary truths to maintain political harmony.
I read a piece in Christianity Today about the "racial retrenchment" within the evangelical church, noting how the "racial reckoning" of 2020 has faded into apathy. The author, Tryce Prince, suggests that the most loving thing one can do is tell the truth, even when it is uncomfortable. The same applies to the state. It is not "pro-business" to lie to the public about the safety of AI; it is a dereliction of duty. We have seen this pattern before, from the urgent creation of agencies after the September 11 attacks to the desperate attempts to coordinate the 2015 European migrant crisis. In every instance, the cost of the delay was paid in human lives.
The trajectory here is clear. Either Vice President Vance leverages his position to persuade President Trump that AI safety is a matter of national security—the only language the current administration truly speaks—or we wait for the "worst-case scenarios" predicted for early 2027 to force our hand.
If we wait for the catastrophe to build the regulator, we have already lost. Legitimacy is a capital stock that spends fast and refills slowly, and we are currently spending it all on the gamble that the machines will be benevolent. The state must either lead this mobilization with the authority of a Manhattan Project or admit that it has abdicated its primary purpose: the protection of the governed. There is no middle ground between a governed system and a collapsed one.