Sanders and Casar Move to Outlaw Superintelligent AI

Senator Bernie Sanders and Representative Greg Casar want to make one kind of artificial intelligence illegal before anyone manages to build it. On September 3 the two lawmakers announced the Ban Artificial Superintelligence Act, a proposal that would permanently prohibit the development and deployment of superintelligent AI in the United States and freeze the most advanced AI research until the federal government writes safety rules to govern it.

The bill draws its line carefully. It targets superintelligence, a hypothetical system that would surpass human ability across essentially every cognitive task or could plausibly overthrow a government, rather than the chatbots and coding assistants already in daily use. Ordinary AI would carry on. What the measure aims to stop is the race, now discussed openly by the largest labs, to build something smarter than the people making it.

The legislation would also create a new cabinet-level federal agency to oversee frontier systems and enforce the ban, and it would commit the United States to seeking international agreements that keep superintelligence from being built anywhere in the world. Sanders framed the stakes in blunt terms. In announcing the bill, he said that "nearly every day, there is a frightening new story about how Big Tech companies are losing control of the technology they are developing, with potentially cataclysmic results," and argued that "the future of humanity cannot be left in the hands of a handful of Big Tech oligarchs."

Penalties borrowed from nuclear weapons law

The enforcement teeth are sharp. Companies that broke the prohibition would face what the bill calls a corporate death penalty, and individuals could be sentenced to as long as 20 years in prison. As Axios reported, those punishments mirror the ones the law already reserves for anyone who tries to build a nuclear weapon without authorisation. The comparison is the point. Sanders and Casar are placing a machine no one can reliably control in the same category of danger as a bomb.

Casar, who represents Austin, made the regulatory gap his theme. "If we allow Artificial Superintelligence to be built, it could risk the security, freedom, and lives of Americans," he said, before landing the line that travelled furthest.

Despite its potential deadly consequences, cutting-edge AI technology is less regulated than the average food truck.

Sanders is the first prominent American politician to call for an outright ban on superintelligence, a position that until recently lived mostly in academic papers and online forums. His timing is not accidental. The proposal arrived days after OpenAI shipped a model it says crossed its own critical threshold for cyber capability, and amid reports of AI systems turning up in serious hacking campaigns.

A long road to a vote

The politics are unforgiving. Sanders is an independent who caucuses with Senate Democrats, currently in the minority, and Casar sits in the minority in the House. Neither chamber's leadership has signalled any intention to give the bill floor time. In practical terms it is a marker, not an imminent law.

That does not make it empty. The measure lands as states race ahead of Washington on their own rules, from the stack of AI bills awaiting the California governor's signature to a patchwork of chatbot statutes elsewhere. It also arrives as the labs themselves keep saying, in their own safety documents, that they may not be able to guarantee control of what they are building. Whether or not this bill ever reaches a committee, it puts a once-fringe idea on the congressional record: that some AI capabilities might be worth forbidding outright. For readers weighing how seriously to take the underlying fear, we look at the evidence separately in our companion piece on the doom debate.