TL;DR — Key Takeaways
– The House passed the bipartisan Ratepayer Protection Act 417-3, directing state utility regulators to consider whether data centers should bear the full cost of new power generation and transmission infrastructure needed to serve them.
– Public concern over AI infrastructure is rising, with nearly two-thirds of Americans expressing serious concern about data centers increasing electricity costs and 57% concerned about their impact on water supplies.
– Lawmakers are also considering a bipartisan proposal that would give AI companies a limited antitrust safe harbor to share threat intelligence and coordinate responses to national security risks such as cyberattacks and illicit model distillation.
In a flurry of legislative action preceding midterm elections, Congress is grappling with the rapid rise of artificial intelligence (AI). But in its haste to regulate AI’s immense power consumption, national security implications, and potential existential risks, it is stuck in a partisan quagmire.
Front and center, before departing Washington, D.C., the House of Representatives overwhelmingly passed a bipartisan bill addressing mounting public anxiety over AI data centers and their strain on the national power grid.
The measure, which passed in a 417-3 vote, directs state utility regulators to consider standards requiring data center operators to cover the full cost of the new power generation and transmission upgrades required to serve them.
Supporters framed the bill as a crucial first step toward shielding consumers from skyrocketing electricity costs. “I promise that this bill is not the end of our focus on these issues. It’s only the beginning,” said Rep. Frank Pallone (D-N.J.), ranking member of the House Energy and Commerce Committee.
The vote comes amid growing voter concern over the physical infrastructure powering the AI boom. According to a recent poll from The Associated Press-NORC Center for Public Affairs Research and the Energy Policy Institute at the University of Chicago, nearly two-thirds of Americans are extremely or very concerned about data centers driving up energy prices, while 57% expressed serious concern over their impact on local water supplies.
Despite the bipartisan consensus on utility costs, deeper rifts remain.
Some lawmakers have called for a total moratorium on new data center construction until comprehensive safety and environmental safeguards are enacted. House Speaker Mike Johnson (R-La.), however, praised the current bill for striking “the right balance of protecting local communities while keeping America on the cutting edge of AI development.”
Meanwhile, hidden within the annual defense policy debates, lawmakers are quietly tackling the national security threats posed by foreign adversaries targeting American AI models. U.S. senators included a targeted antitrust exemption for AI companies inside a manager’s package of amendments for the National Defense Authorization Act (NDAA).
Authored by Sens. Jim Banks (R-Ind.) and Adam Schiff (D-Calif.), the provision would grant AI firms limited immunity to share threat intelligence and coordinate defenses specifically against Chinese espionage, cybersecurity incidents, and “model distillation,” a practice where foreign actors train lower-tier models on advanced American systems to steal proprietary techniques.
While the narrow exemption drew support from tech giants like Google, broader legislative progress on the $1.5 trillion defense budget remains stalled. Senate Democrats blocked debate on the measure over the summer due to disputes surrounding President Donald Trump’s handling of ongoing foreign conflicts. Analysts at the Center for Strategic and International Studies do not expect a finalized defense bill to reach the president’s desk until at least December.
The national security waiver arrives amid a fierce broader debate over tech regulation. Anthropic CEO Dario Amodei and other industry leaders have advocated for wider antitrust exemptions that would allow companies to coordinate on pacing technology rollouts. However, that proposal faces sharp opposition from both Capitol Hill and former administration officials who argue safety should not be conditioned on regulatory waivers.
The Senate floor also saw direct drama over AI safety precautions. Sen. John Kennedy (R-La.) sought unanimous consent to pass a bill requiring tech developers to build a mandatory kill switch into superintelligent AI systems, enabling human operators to shut down models that pose catastrophic risks.
Kennedy’s effort was abruptly blocked by Sen. Rand Paul (R-Ky.), who argued that an open-ended federal mandate would stifle innovation and disrupt the broader economy.
Paul insisted the complex technology requires deeper study before Congress imposes sweeping hardware or software requirements, leaving Capitol Hill deeply divided on how quickly to rein in the evolving technology.
“You can’t control how the political and financial battle lands, and you shouldn’t try. But you can decide today that your organization’s memory is yours – legible, held in a repository you own, portable to any model, governed by you,” said Chris Kincade, CEO of Starling Memory Works. “That’s an architecture decision, not a policy outcome. It doesn’t need Washington to legislate, Wall Street to fund it, or the labs to agree. The organizations that make it now are the ones still standing whichever way the fight goes – because their intelligence was never anyone else’s to lose.”
Still, Treasury Secretary Scott Bessent urged lawmakers on Tuesday to support the development of open-source AI, arguing that open models are essential for the U.S. to maintain its competitive edge against Chinese rivals.
Testifying before the House Financial Services Committee during a hearing on the economy and international finance, Bessent highlighted AI as a critical strategic priority. He pointed to Anthropic’s Mythos model, released earlier this year, as a key milestone toward artificial general intelligence, while cautioning that concentrating AI power among a few dominant firms could hamper American progress.
“One of the things I would say is that we need to develop… more open-source models in the U.S.,” Bessent told the panel. “We can’t let these large labs have regulatory capture because that will stop innovation.” His remarks underscore growing concern within the administration that strict industry oversight or monopolistic control could stifle smaller developers and tip the balance of global AI leadership toward China.
In a viral speech addressed directly to industry leaders, actor Joseph Gordon-Levitt urged the public to turn out at the ballot box this November to force government regulation of artificial intelligence, warning that “AI companies are operating in such a way that the law should be shutting them down.” His impassioned call to action arrives amid mounting alarm over the technology’s expanding influence and growing demands from experts for a coordinated pause on AI development.
“Any organization that publishes an AI governance framework, whether to satisfy (California Senate Bill) 53, the EU AI Act, ISO 42001 or a customer contract, is making promises about its own behavior that can be held against it later,” said Patrick Sullivan, vice president of strategy and innovation at A-LIGN. “Promise only what you can prove you did, and create the paper trail as you make each decision, before, not after, someone comes asking.”
