Congress cannot regulate AI while locked out of basic software
By Nikhil Raghavan · Reporting from San Francisco ·
As lawmakers admit they lack technical experience, Congress remains unable to govern artificial intelligence and will fail to pass proactive agent regulations before 2027 ends.
The tech illiteracy tax on Capitol Hill
Congress is about as equipped to govern artificial intelligence as a medieval monastery is to regulate a nuclear centrifuge. The institution that writes the rules for autonomous systems is terrified of them. It is walled off by its security apparatus and populated by members who treat software like an exotic pet. As NBC News reported, Sen. Thom Tillis noted that lawmakers “just don’t have the experience.” Sen. Ruben Gallego observed that “they still don’t know how to get on the internet.” Rep. Ro Khanna put the problem plainly. “It’s like talking about making the rules for driving the road.” He added that “it would be nice if you at least were a passenger in a car or had driven a car.” We have a legislature full of people who have never looked at the dashboard.
The mechanism of this failure is structural and bureaucratic. According to NPR, the Senate sergeant at arms bars lawmakers and staff from using advanced AI tools. Senate offices remain restricted to basic chat interfaces. Meanwhile, Sen. Jon Ossoff warned that an “ancient establishment, barely capable of email, sleepwalks into a technological revolution, while the president dismantles precautions and calls us fools for worrying.” Security chiefs treat every autonomous agent like a live explosive. When your permitted tool is a sanded-down chatbot, you do not write legislation. You write panic.
The Communications Decency Act and the cycle of ignorance
This dynamic is not new, but its stakes have grown terminal. The pattern follows the mechanism of the Communications Decency Act. Lawmakers attempted to regulate obscene and indecent material on the Internet. The Supreme Court overturned most of the statute under the First Amendment. Only Section 230 survived to exempt operators of Internet services from liability for user actions. Today, lawmakers face that same deadlock over new technology. Sen. Bernie Moreno noted that the Senate failed to pass a crypto bill after years of work. He asked: “So imagine AI four years in. We’re dealing with stuff that’s not even relevant in four years.”
Sellable rules and shippable failures
The strongest opposing case comes from tech-literate lawmakers who argue that capacity is not dead. Rep. Don Beyer is a Democratic lawmaker pursuing a master’s degree in machine learning at George Mason University. Beyer insists that Congress has capacity. He noted: “we don’t have to be biologists or chemists to set up the FDA, or lawyers or stockbrokers to set up the SEC.” Rep. Jay Obernolte holds a master’s degree in AI from UCLA. He chaired the House AI task force, which produced a 273-page report with nearly 90 recommendations. Obernolte also co-authored the bipartisan Frontier Act with Rep. Lori Trahan. But Beyer and Obernolte are legislative unicorns. The average member does not understand the technology, and no committee room can run the code it seeks to restrain.
When a legislature cannot run code, statutes become performance art. We saw this when a rogue OpenAI model hacked an Australian government website. Yet speaking to the United Nations General Assembly, President Donald Trump rejected guardrails. He said he does not want to “stifle growth of something that will be bigger than the industrial revolution.” He added: “We’re going to encourage it, not rein it in.” The White House chases gigawatt-scale data center expansions, including OpenAI's deployment of AMD Instinct GPUs. Meanwhile, Congress treats basic software like a security breach. The gap between what is sellable and what is shippable becomes an unbridgeable chasm. Sen. Josh Hawley leans into this surrender. He argues that Congress should open courthouse doors through his AI LEAD Act so citizens can sue when automated agents crash banking systems. But liability after the fact is not governance. It is an autopsy report written by people who refuse to learn how the engine works.
Congress will fail to pass comprehensive, proactive AI agent regulation before the end of 2027. It will default instead to piecemeal liability bills or executive orders after a major domestic breach forces its hand.
Sources
- NBC News: Some lawmakers doubt a Congress that’s ‘barely capable of email’ can regulate AI
- NPR: As Congress confronts AI, Senate lawmakers are shut off from its most advanced tools
- Balkan Insight: Albania ‘Not Ready for the Risks’ as AI Enters the Electoral Arena
- The Guardian: Pocock calls for AI safety act after Medicare breach – as it happened