
Share
Scott Wiener built his political career on AI safety bills in California. Now headed to Congress, he faces a Washington gridlocked on tech policy, and a public already furious about where AI's infrastructure actually plugs in.
Nancy Pelosi represented San Francisco in Congress for nearly four decades. Her likely successor spent years in Sacramento trying to put guardrails on the technology industry that calls her old district home.
Scott Wiener, the California state senator who authored some of the most ambitious AI safety legislation in the country, has won the race for Pelosi's open congressional seat. He's heading to Washington with AI risk already on his mind, and he's not being subtle about it. In a 20-minute conversation just hours after his victory was declared, he used the phrase "recursive self-improvement" six times. That's the term researchers use for a hypothetical scenario where an AI system gets good enough to improve its own design, potentially faster than humans can monitor or control.
"We need to slow this down," Wiener said. "Because it's moving very quickly, and we could sleepwalk into a disaster." It's a striking line from someone about to join a federal legislature that has, so far, struggled to pass much of anything on AI governance. Wiener himself isn't pretending otherwise. He describes himself as only "cautiously optimistic" about Congress's ability to actually rein in the technology once he gets there.
That caution is worth sitting with. State legislatures, California included, have moved faster than Congress on AI oversight largely because federal lawmakers remain split on basic questions: whether to regulate at the state or federal level, how to balance innovation against safety, and how to handle an industry that insists any serious guardrail will hand the race to China. Wiener's own track record in Sacramento shows what's possible when a legislator pushes hard on frontier AI risk, but it also shows how fiercely the industry pushes back. Whether that translates into federal wins is a different test entirely, and one Wiener will be taking on almost immediately.
While Wiener was celebrating in San Francisco, a different story about AI's costs was unfolding roughly 2,800 miles away in Ashburn, Virginia. Ashburn sits at the heart of Data Center Alley, the world's largest concentration of data centers, and it's become ground zero for a backlash against AI infrastructure that's spreading across the country.
The tech industry has tried to frame the pushback as a kind of culture war, casting opponents as reflexively anti-progress. But time spent in Ashburn suggests something more specific and more practical is driving the anger: electricity. Communities react very differently depending on whether a data center pulls power from the existing grid, competing with homes and businesses for capacity, or generates its own power onsite, which raises separate concerns about emissions, noise, and land use. Either way, the industry's appetite for power is enormous, and it's not something a better PR campaign can paper over.

This matters because data centers aren't an abstraction anymore. They're neighbors. They show up as higher electricity bills, new transmission lines, and construction noise, and that turns an esoteric debate about AI capability into a very local argument about who bears the cost of the boom. Scott Wiener's "recursive self-improvement" is a long-term, existential framing of AI risk. The Ashburn backlash is the opposite: immediate, physical, and already showing up on utility bills.
Both stories are really about the same underlying tension. The AI industry wants to move fast, and the systems meant to check that speed, whether political oversight in Washington or power grids in Virginia, are straining to keep up. Wiener's bet is that federal legislation can catch up to the technology before something goes badly wrong. The Ashburn story suggests that even if lawmakers never pass a single AI safety bill, the physical limits of electricity supply might end up doing some of that regulating anyway, just not by design.
There's a case to be made that this is actually good news for anyone worried about unchecked AI growth: infrastructure constraints are a kind of natural brake. But it's a crude one. Grid capacity doesn't care whether a given AI model is safe or dangerous, useful or wasteful. It just cares about supply and demand. Relying on electricity shortages to slow AI down means the slowdown will hit good and bad uses alike, and it will hit the communities nearest the data centers hardest, regardless of what Congress eventually decides to do.
Other signals this week point to the same uneasy mix of promise and risk that's defining this moment in AI policy. Australia is weighing criminal penalties for AI agents after one was linked to a government systems hack, an early sign that governments are starting to treat autonomous AI behavior as a matter for criminal law rather than just corporate liability. Meanwhile, two US towns, York, Maine, and League City, Texas, voted to ban Flock surveillance cameras, adding to a growing list of communities pushing back against AI-powered monitoring in public spaces. Separately, reporting this week has raised questions about Nvidia CEO Jensen Huang amid an FBI investigation tied to an alleged affair and espionage allegations, with analysts warning the controversy could trigger new chip export restrictions and complicate an already fraught relationship between the chip industry and Washington.
Wiener's arrival in Congress and the Ashburn backlash are two sides of the same unfinished argument about who gets to decide how fast AI moves, and who pays the price along the way. Lawmakers are still debating existential risks that may take years to materialize, while communities near data centers are already living with the consequences of AI's physical footprint today. Getting federal policy right will mean addressing both timelines at once: the slow-moving danger of systems that could outpace human oversight, and the immediate, tangible cost of powering them. Neither problem will wait for the other to get solved first.
Tags
Original Sources
The Download: Pelosi’s successor targets AI and data centers face power woes
↗ https://www.technologyreview.com/2026/11/10/1148150/the-download-pelosi-successor-ai-regulation-data-center-backlash
About the author
Amara's entry point into AI was an epidemiology role at a London research hospital, where she spent five years studying how digital health tools reached — or conspicuously failed to reach — underserved communities. Watching early algorithmic systems in healthcare quietly entrench existing inequalities, she redirected her career toward the systemic consequences of AI at scale. She covers AI through an unflinching lens: who benefits, who bears the cost, and what evidence actually says versus what the press release claims. Her writing is calm and precise, but she doesn't mistake balance for neutrality.
More from The Steward →This Week's Edition
11 November 2026
14 articles
Related Articles

Europe's Digital ID Push Tests a New Privacy Tool: Zero-Knowledge Proofs
Policy & Regulation · 5 min

Why the Woman Who Fired Sam Altman Still Doesn't Trust AI's Boardrooms
Policy & Regulation · 5 min

Pakistan to Send Climate Minister to India for First High-Level Visit in Three Years
Policy & Regulation · 5 min
Related Articles

Europe's Digital ID Push Tests a New Privacy Tool: Zero-Knowledge Proofs
Policy & Regulation · 5 min

Why the Woman Who Fired Sam Altman Still Doesn't Trust AI's Boardrooms
Policy & Regulation · 5 min

Pakistan to Send Climate Minister to India for First High-Level Visit in Three Years
Policy & Regulation · 5 min
More Stories
© 2026 Cedar & Bloom. All rights reserved.