Authored by Jordan Schachtel via American Greatness,
The AI Kill Switch Act, introduced Thursday by Rep. Ted Lieu (D-Calif.) and Rep. Nathaniel Moran (R-Texas), seeks to hand the government unprecedented authority to order private companies to throttle, suspend, or shut down products in "kill switch" fashion, purely on their own arbitrary determination, with fines up to $20 million a day. Behind this carefully staged rollout is a fanatical coalition that looks like a broad consensus, but it is mostly one funding network with an arsenal of astroturf organizations at its disposal.
The Lieu-Moran bill arrived with endorsements preloaded from five groups: the AI Policy Network, Americans for Responsible Innovation, ControlAI, the Future of Life Institute, and the Alliance for Secure AI. Five organizations may sound like a broad coalition, so let's pull the thread and explore the details.
Americans for Responsible Innovation is funded by Coefficient Giving. Coefficient Giving is the freshly rebranded Open Philanthropy, the vehicle for Facebook billionaire Dustin Moskovitz and Cari Tuna's fortune and the central bank of the far-left Effective Altruism (EA) movement's AI doom apparatus. Dustin Moskovitz was one of the largest individual megadonors supporting Kamala Harris's 2024 presidential election effort, contributing roughly $38 million to the pro-Harris super PAC Future Forward (FF PAC). The Open Philanthropy Action Fund pumped $2.9 million into ARI. ARI's 501(c)(3) sister organization collected another $3 million in general support.
The Alliance for Secure AI is the group built to make all this palatable to the Right. It spent six figures on ads running on Fox News and Newsmax last summer, pitching AI fear to conservative audiences. Who pays for the conservative outreach? In May, the New York Times reported that the Alliance for Secure AI is funded by ... Coefficient Giving. Surprise! You will search its website for that disclosure in vain.
ControlAI, the third member, openly advocates an international prohibition on advanced AI development and swims in the same donor waters. The group's top leaders have been calling for a complete shutdown of AI development since the very early iterations of ChatGPT.
The Future of Life Institute, which has also received millions from Coefficient Giving, is an EA-aligned leftist nonprofit best known for its ruthless insistence that America must force industry to "pause AI" indefinitely.
The AI Policy Network (AIPN) is an AI doomer outfit that is supported by leftist tech bros in Silicon Valley, some of whom are longtime AI panic mongers and influential executives at Anthropic and OpenAI.
This power grab arrives on the heels of the OpenAI sandbox episode, in which the company itself reported a security breach that has not been independently verified. OpenAI has every incentive to inflate the dangers of its own product because a frightening "autonomous" AI is both a better marketing pitch and a better argument for the regulatory moat that keeps its future competitors out. Washington is now being asked to hand a cabinet secretary sweeping authority over private enterprise on the strength of an incident report written by the party that stands to benefit.
So the "broad coalition" is substantially one checkbook, belonging to a left-wing Silicon Valley billionaire, laundered through a series of organizations that act as window dressing for one agenda. This is not bipartisanship, but rather the lobbyist-concocted appearance of bipartisanship.
The bill empowers the Homeland Security secretary with a shutdown trigger that includes a loss-of-control scenario, defined to cover a model behaving contrary to instructions in "another high-stakes context." That phrase is defined nowhere. The company must comply immediately without recourse. Its petition for reconsideration, the statute says flatly, does not stay the order. If the secretary ignores the petition for five days, silence counts as a denial. Court review would, of course, come months later, long after the product went dark, the business was likely destroyed, and all of the customers went elsewhere. Do we really want to grant the government this power?
And the definitions of who is covered? Rewritten by the agency, by rule, every single year, guided by a factor list that ends with "such other factors as the secretary determines relevant." Try executing a successful business strategy when your legal exposure gets redefined annually by government bureaucrats.
Some frontier AI model companies are not yet protesting the bill, largely because this enormous compliance burden could potentially create a wall around the incumbents. The giants can absorb the lawyers, the audits, the whims, and the like. Startups that attempt to catch up could get obliterated in one fell swoop. This kind of regulation is deliberately priced at a level only the largest firms can bear. It is a moat with a safety label.
The AI doomer lobby wants its federal chokepoint, and some of its frontier lab collaborators get a licensing wall between themselves and their future competitors. Everyone in this somewhat complex coalition wins except the American innovators.
What Congress is being sold is a federal "kill switch" button for upstart private enterprise, wielded on undefined standards and immune to timely challenges; and it's all being pushed by a fake "movement" that is mostly one billionaire's philanthropic arm.
Jordan Schachtel is an independent journalist who covers AI, tech, and world politics, and he's the publisher of The Dossier on Substack.