Proposed Superintelligence AI Ban
Lawmakers Introduce Landmark Bill to Ban Superintelligent Artificial Intelligence Systems
Anonymous
Key Points
1
The proposed rules would immediately pause development of advanced AI systems trained with at least 10^25 floating-point operations, a threshold intended to capture frontier models driving the current AI race.
2
Under the proposal, AI laboratories would have 24 hours to report suspected superintelligence or precursor capabilities, and any system classified as artificial superintelligence would have to be rendered inoperative within 30 days.
3
Developers would need a government-issued charter to build or distribute covered advanced models, while companies would require government approval before deploying them.
4
The bill would also target dangerous capabilities that could emerge before a system reaches full superintelligence, including attempts to evade oversight and certain forms of unauthorized access to computer systems.
5
Sanders said the risks extend beyond human extinction to financial, defense, energy and healthcare systems, as well as “virtually every computer that is connected to the internet.”
Sen. Bernie Sanders and Rep. Greg Casar introduced the Ban Artificial Superintelligence Act on September 23, 2026, proposing to permanently prohibit AI systems that surpass human intelligence or could threaten humanity News Sources. The bill would immediately pause development of the most advanced AI systems, establish a new cabinet-level Department of Artificial Intelligence, and impose prison sentences up to 20 years for violations News Sources.
The legislation faces steep odds in the Republican-controlled Congress, as President Donald Trump favors accelerating AI development to compete with China News Sources. Yet tech workers at major AI companies and prominent researchers including Geoffrey Hinton have publicly backed the measure, citing catastrophic risks to critical infrastructure and mass job displacement News Sources.
What the Bill Would Actually Do
The act targets AI systems trained with at least 10^25 floating-point operations—a threshold meant to capture frontier models now driving the race News Sources. Labs would have 24 hours to report suspected superintelligence. Any system classified as superintelligent must be shut down within 30 days News Sources.
Companies wanting to build or deploy advanced AI would need a government charter News Sources. The bill also bans recursive self-improvement—where AI systems automatically make themselves smarter—and blocks using AI to design nuclear or biological weapons News Sources.
Sanders and Casar Frame It as a Public Safety Brake
Sanders warned of threats far beyond extinction: financial systems, hospitals, power grids, and nearly all connected computers at risk News Sources. "When you are racing towards a cliff, you don't just ease up on the gas pedal," Sanders said. "You hit the brakes."
Rep. Casar, Chair of the Congressional Progressive Caucus, targeted the core concern: "Experts are warning that AI superintelligence, wielded by the wrong humans or by rogue AI, could kill countless numbers of people" News Sources. He called out Trump's opposite stance, saying "Something has to change."
Trump and Republicans Oppose the Ban as Economic Suicide
President Trump and GOP lawmakers frame the bill as anti-innovation overreach that would weaken U.S. national defense News Sources. They argue slowing American AI development hands technological dominance to China, undermining critical military and economic advantages News Sources.
The measure would temporarily freeze hundreds of billions in capital investments in frontier AI data centers and shift corporate focus toward safety alignment News Sources. Conservative leaders cite energy grid strain and regulatory burden as additional reasons to reject federal restrictions on advanced AI development News Sources.
A Technical Problem: Vague Definitions Could Fail Enforcement
The bill does not set clear technical thresholds for determining when a system qualifies as superintelligent News Sources. Critics worry that vague legal definitions of "capabilities matching human cognition" create enforcement traps and could let foreign rivals catch up News Sources.
Tech workers and safety researchers support the intent but face uncertainty about how agencies would measure superintelligence in practice News Sources. The measure also directs the State Department to negotiate binding international treaties banning superintelligence research globally—a goal most AI-leading nations are unlikely to accept News Sources.
Sentiment
Mostly NegativeNEGATIVENEUTRALPOSITIVE
Publishers
28
Articles
201
Reach
229
A
AnonymousAuthor
