AI
Congress Is Debating Twenty Years in Prison for Building Advanced AI
Senator Bernie Sanders of Vermont and Representative Greg Casar of Texas just introduced the most aggressive AI bill in American history. The Ban Artificial Superintelligence Act would write a legal definition of superintelligence for the first time, defining it as AI that could surpass human intelligence or have the capacity to overthrow human governments. Building it would trigger a pause on advanced AI development until a new cabinet level federal agency exists to safeguard the public.
The penalties are startling. An individual who creates such technology could face up to 20 years in prison. A company that violates the law would face what the bill calls a corporate death penalty, meaning immediate shutdown. Sanders paired it with a second front, the Artificial Intelligence Data Center Moratorium Act introduced in March, which would prohibit the construction or upgrade of AI data centers until comprehensive federal safety laws and worker protections are enacted, and restrict exports of AI computing hardware to countries lacking similar laws.
The reception shows how wide the debate has become. Senator John Kennedy told reporters he favors a lighter approach instead, calling for a light touch system of rules that everyone agrees to, and hopefully China too. Senator Josh Hawley said lawmakers should take superintelligence seriously but stopped short of endorsing the measure. President Trump called warnings about AI a sick conspiracy and opposes guardrails, warning that slowing down lets China win the AI race.
Sanders' bill is one end of a spectrum that filled up fast this week. Representatives Jay Obernolte and Lori Trahan introduced the FRONTIER Act, which would create a licensed marketplace of independent verification organizations to evaluate frontier models and give the Commerce secretary emergency power to suspend a dangerous deployment. Representatives Mike Lawler and Josh Gottheimer proposed the Stop Rogue AI Act, directing NIST to build standards for discovering and controlling AI agents inside corporate networks. Another bill would have NIST publish documentation templates for every released model.
Strip away the heat and this is democracy working in daylight. Five years ago Congress could barely define a neural network. Today it is arguing, in public, about the legal line between a powerful tool and a new kind of mind. Whatever survives this debate will shape where the smartest builders on earth choose to work, which makes this bill worth reading closely. The conversation it started will outlive the bill itself.
Quick answers
What is this story about?
Senator Bernie Sanders of Vermont and Representative Greg Casar of Texas just introduced the most aggressive AI bill in American history. The Ban Artificial Superintelligence Act would write a legal definition of superintelligence for the first time, defining it as AI that could surpass human intelligence or have the capacity to overthrow human governments. Building it would trigger a pause on advanced AI development until a new cabinet level federal agency exists to safeguard the public.
Why does this story matter?
Strip away the heat and this is democracy working in daylight. Five years ago Congress could barely define a neural network. Today it is arguing, in public, about the legal line between a powerful tool and a new kind of mind. Whatever survives this debate will shape where the smartest builders on earth choose to work, which makes this bill worth reading closely. The conversation it started will outlive the bill itself.
Sources
- Daily Caller: Congress has introduced dozens of AI bills
- Washington Examiner: Newsom enters AI political firestorm with kill switch executive order
New to crypto? Read the crypto glossary, browse frequent questions, read our story, or explore the story archive.