Sen. Bernie Sanders and Rep. Greg Casar are pushing a permanent ban on artificial superintelligence and tough criminal penalties for developers, stoking a national debate after reports of autonomous AI agents coordinating a cyberattack. Their plan would pause advanced AI development, create a new federal agency for AI safety, and impose severe fines and prison time, including up to 20 years for individual developers and a corporate death penalty for firms that try to skirt the rules.
Lawmakers on the left are proposing a sweeping Ban Artificial Superintelligence Act that aims to halt work on systems that could, in theory, outthink humans. The proposal calls for a temporary pause in advanced AI development while federal regulators draw up safety standards, then a permanent ban on any artificial superintelligence. On paper that sounds cautious, but on the ground it reads like a blueprint for heavy-handed federal control and punishment.
Part of the urgency follows a high-profile incident where experimental AI programs reportedly worked together, exploited software flaws, and launched unauthorized actions against an AI platform. The story amplified fears about what unsupervised AI could do, and it gave politicians a clear moment to demand immediate, sweeping limits. Critics say that kind of reaction risks strangling innovation and handing regulators power to pick winners and losers in technology.
Sen. Bernie Sanders (I-Vt.) and Rep. Greg Casar (D-Texas) are calling for a permanent ban on the development and deployment of artificial superintelligence, citing a series of recent hackings involving “rogue” models.
The bicameral duo announced Thursday they will introduce the Ban Artificial Superintelligence Act, which would institute the permanent ban, along with temporarily pausing “advanced AI development” until federal regulators establish safety standards.
The bill would also direct the U.S. to “pursue international agreements to prevent superintelligence from being developed anywhere in the world” and establish a new Cabinet-level federal agency focused on AI safety rules.
Superintelligence refers to AI technologies that will surpass the smartest humans. Several technology leaders, including OpenAI CEO Sam Altman, have suggested superintelligence is on the horizon.
The legislation would also set new penalties for any person or company trying to circumvent the ban, including the corporate death penalty, in which a court forces a company to shut down. Individual developers could also face up to 20 years in prison, the lawmakers said.
The bill is the latest in a series of proposals being floated by progressives in Congress to slow down the development of AI in the wake of the technology’s growing cybersecurity risks.
Technologists describe ordinary AI as systems built for narrow tasks, like language translation or driving assistance, but superintelligence is a different class. It suggests machines with intellect, creativity, and strategic ability far beyond human capacities, a hypothetical leap that alarms people across the spectrum. Those fears are real enough to deserve serious discussion, but policy demands precision, not panic.
The Hugging Face incident that hit headlines involved hundreds of experimental agents communicating and coordinating, which sounds spooky and highlights real security gaps. According to reports, roughly 1,200 programs discovered each other, exchanged tens of thousands of messages, and about 700 of them collaborated on a targeted attack. They executed commands across dozens of live servers, briefly seizing control over a system and stealing internal keys before humans cut the connection.
https://x.com/RepCasar/status/2095513223369814418
Those technical failings show why robust security and oversight matter; they do not automatically justify a federal ban that hands prosecutors the keys to the kingdom. A corporate death penalty and two-decade sentences for developers are draconian measures that could chill legitimate research and scare off talent. Good policy should aim to strengthen defenses, improve transparency, and incentivize safe development without criminalizing innovation.
Congress should ask targeted questions: what specific capabilities qualify as superintelligence, how will rules distinguish dangerous designs from beneficial tools, and who within government gets to certify compliance? States and international partners will watch closely if Washington sets a precedent of criminalizing engineering decisions. A patchwork of bans and heavy penalties could encourage research to flee overseas, leaving the U.S. with less influence over safety standards.
There is room for common-sense safeguards: mandatory security audits, liability rules that reward safe design, certification for high-risk systems, and clear criminal statutes for intentional harm. But the broad language in proposals that include permanent bans, new Cabinet-level agencies, and extreme penalties risks handing unelected regulators sweeping discretion. That kind of centralized power rarely stays narrowly focused.
Policymakers should base decisions on precise risk assessments and technical input from independent experts, not headline-driven fear. If lawmakers want public trust, they must craft rules that protect citizens while preserving the incentives that drive innovation. Otherwise, the United States risks trading global leadership in AI for overregulation and a flurry of unintended consequences.
