Sanders and Casar Unveil a Bill to Ban Superintelligence — With Prison Terms Up to 20 Years

A sweeping AI bill landed in Congress on Wednesday with a number attached: 20. Sen. Bernie Sanders and Rep. Greg Casar unveiled legislation that would permanently ban artificial superintelligence, pause development of the most advanced AI systems, and route enforcement through a new Cabinet-level agency — the Department of Artificial Intelligence — with violations carrying up to 20 years in prison in some cases. The bill was provided first to the Associated Press (AP).

The timing does half the analysis. On Tuesday, President Trump told the UN General Assembly that US documents will call AI "super intelligence" from now on. On Wednesday, Sanders introduced a bill that would criminalize developing the thing that new name describes. "It doesn't take a genius to say, 'slow it down,'" Sanders, I-Vt., told the AP. "Do we really want to develop a super intelligence that when it becomes smarter than human beings could act independently of human control? I don't think we do."

What Actually Happened

The legislation from Sanders and Casar — a Texas Democrat who chairs the Congressional Progressive Caucus — has four load-bearing pieces, consistent across the AP's report and Fox Business's coverage of the legislative framework released by Sanders' office:

• A permanent prohibition on artificial superintelligence, defined as a system that exceeds human cognitive ability or has sufficient capabilities to plan and execute the destruction or disempowerment of humanity (AP). Fox Business reports the definition also covers systems with dangerous capabilities including the ability to subvert shutdown commands.
• A temporary pause on development of the most advanced AI systems until the new Department of Artificial Intelligence establishes federal safety rules and a model review process.
• A pre-deployment gate: advanced AI systems would need federal approval before they could be deployed.
• Enforcement teeth: penalties for attempting to violate or circumvent the restrictions, with individuals facing up to 20 years in prison in some cases.

The bill would also make it US policy to pursue international agreements and coordination with allies aimed at preventing superintelligence development around the world (Fox Business). Sanders framed the challenge as Cold War arms control: "Ronald Reagan, arch-conservative, was smart enough to understand that he had to sit down with Gorbachev, a communist, and work out a nuclear treaty that protected humanity," he told the AP. "I think we can do that again now." Last week, Sanders publicly urged Trump to negotiate with Chinese President Xi Jinping on a potential pause in advanced AI development and a ban on superintelligence (Fox Business) — and AI safety is expected to be a central part of Trump-Xi talks at the White House this week as the countries compete for dominance in the sector (AP).

The Employee Statement Is the Real Signal

Bills from the progressive wing face long odds in this Congress. Signed statements from inside the labs are a different kind of event.

A number of current employees at leading AI companies endorsed the bill in a statement shared with the AP. Juan Felipe Cerón Uribe, a researcher in OpenAI's Safety Systems group, said "superintelligence could either go extremely right or extremely wrong," adding that society "shouldn't be playing such games." Swante Scholz, a software engineer at Google DeepMind who said he was not speaking on behalf of his employer, went further: on the current development path, "the most likely outcome is an existential catastrophe for humanity," and "a ban on superintelligence development would be a positive change for the foreseeable future."

The employers are not on the letter. Fox Business says it reached out to OpenAI and Google DeepMind for comment; neither company has endorsed anything close to a development ban. That gap — named employees calling for a criminal ban on the frontier while their employers race toward it — is now a matter of public record. Sanders walked with Geoffrey Hinton and Max Tegmark ahead of an AI briefing for senators on Sept. 16 (Fox Business). The safety movement's bench is no longer outside the industry; it is inside it, on the payroll, with quotes attached.

Long Odds, Real Positioning

The AP's own assessment: the bill faces long odds in the Republican-controlled Congress, where lawmakers have struggled to coalesce around even less sweeping AI regulations. Trump has resisted recent calls for greater government oversight — putting him at odds with tech leaders including Anthropic's Dario Amodei, OpenAI's Sam Altman and xAI's Elon Musk — and told the UN on Tuesday that the Justice Department could step in if necessary to rein in the industry (AP; Fox Business).

But the bill is a marker with a calendar attached. November's midterms are six weeks away, and the AP reads the legislation as a window into how the progressive wing of the Democratic Party is approaching AI regulation ahead of those midterms and a 2028 presidential campaign in which the technology is expected to be increasingly prominent. The lane is being paved at the state level too: Maryland Gov. Wes Moore announced a plan to regulate AI on Tuesday, and California Gov. Gavin Newsom signed an executive order last week to accelerate implementation of a California law calling for independent oversight of AI companies (AP) — with Newsom's office announcing a roster of experts on Wednesday to deliver on the order, including work on a frontier AI "kill switch" (CA.gov).

Sanders, 85 and a two-time presidential candidate, has been building toward this for months: a June bill to create a sovereign wealth fund financed through a one-time tax on the stock of the largest AI companies, a briefing for colleagues with experts last week, and a conference appearance alongside Republicans including former White House adviser Steve Bannon (AP). The traditional left-right map does not fit this issue. Polling from the AP-NORC Center and the University of Chicago's Energy Policy Institute finds 53% of Americans "extremely" or "very" concerned about AI's environmental impacts — up from 41% a year ago — with Democrats driving much of the increase.

The counterargument that has killed every pause proposal to date — slowing the United States while China keeps advancing — remains the bill's biggest obstacle (AP). Sanders' answer is to make the ban itself the negotiating position: you cannot fence off superintelligence in one country, so put a global ban on the table and talk to Beijing from there.

What Builders Should Take From It

• Pre-deployment approval is now on the record as a live federal proposal. Even at long odds, the compliance surface is being drafted from both directions — federal review in Congress, independent oversight in California. Audit trails, evals, and model documentation are sellable in every scenario that lands.
• "Superintelligence" is now a legal battleground, not just a talking point. One branch of government adopted the term as branding on Tuesday; the next day a bill defined it as a system that exceeds human cognition or defeats shutdown commands and proposed to criminalize it. How vendors map their safety claims to that definition will matter in procurement.
• Insider dissent is now quotable. Named OpenAI and DeepMind employees calling for a frontier ban will surface in enterprise due-diligence conversations and competitor sales decks. Plan for the question.
• The arms-control frame has entered mainstream US politics. With Trump-Xi talks this week and Sanders explicitly invoking Reagan and Gorbachev, international-coordination language is coming from both parties. Expect "AI treaty" vocabulary to reach standards bodies and RFPs.
• Price this as signal, not regulation. Nothing here becomes law this Congress. The near-term product is positioning for midterms and 2028 — which shapes state laws, procurement templates, and buyer sentiment long before any vote is taken.

Developer312 covers the AI business signals builders actually need to act on. Get the weekday briefing at developer312.com.