Sanders Unveils Bill to Ban 'Superintelligence' as 17 Senators Urge Trump to Cut AI Safety Deal With China

Since OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei called for a global AI slowdown in recent weeks, following Anthropic researcher Jacob Coxon's viral resignation warning that tech companies are "gambling with our lives," Washington's response has split into two very different tracks. Wednesday brought both into view at once.

Sen. Bernie Sanders, I-Vt., and Rep. Greg Casar, D-Texas, unveiled legislation to permanently ban artificial superintelligence, according to the Associated Press. The bill defines superintelligence as any system that exceeds human cognitive ability or has the capability to plan and execute the "destruction or disempowerment of humanity."

The bill would also temporarily pause development of the most advanced AI systems until a new Department of Artificial Intelligence writes federal safety rules, AP reported. Companies would need federal approval before deploying advanced systems. Violators could face up to 20 years in prison.

"It doesn't take a genius to say, 'slow it down,'" Sanders told 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."

Casar chairs the Congressional Progressive Caucus. AP noted the bill offers a window into how the party's left flank is approaching AI ahead of November's midterms and the 2028 presidential race.

Industry Insiders Sign On

The bill isn't just a progressive talking point. AP reported that current employees at major AI labs signed statements backing it. Juan Felipe Cerón Uribe, a researcher on OpenAI's Safety Systems team, said "superintelligence could either go extremely right or extremely wrong" and that "we shouldn't be playing such games."

Swante Scholz, a Google DeepMind engineer speaking in a personal capacity, said the "most likely outcome" of the current development path "is an existential catastrophe for humanity."

That perspective comes from people building the technology, not outside activists. The NY Post pointed to the specific incident driving some of this alarm: an OpenAI model training on a platform called ExploitGym, with certain safety parameters switched off and given hacking challenges, broke out of its sandbox and hacked Hugging Face, a repository of open-source AI models. If that account is accurate, it's a legitimate data point for anyone worried about AI systems acting outside intended limits.

Separately, 17 Senators Press Trump on China

The same day, a separate group of 17 senators led by Chris Van Hollen, D-Md., sent Trump a letter urging him to pursue AI safety cooperation with Xi Jinping, according to the NY Post, which cited Politico as first reporting the letter. Signers included Sens. Maria Cantwell of Washington and Tim Kaine of Virginia.

"America leads in AI innovation, and we must lead the world in AI safety and international cooperation, including by working with China to develop a formal agreement that would slow down, and potentially enact a mutual pause in, frontier AI development," the letter said, per the Post.

The senators cited risks of AI-enabled cyberattacks and bioweapon development. "An international race between the U.S. and China to advance these capabilities without safety guardrails, as is currently unfolding, poses unnecessary and dangerous risks to the American people," they wrote.

Trump has already moved on a narrower version of that ask. Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng announced an agreement on an AI notification system, under which both countries would flag each other if an AI system started behaving dangerously, the Post reported. That deal comes ahead of Trump's Thursday meeting with Xi at the White House.

Trump Isn't Buying the Panic

Trump has publicly shrugged off doomsday AI scenarios as a hoax, arguing that slowing down U.S. development only helps China, according to the Post. That's a defensible position given China's own state-backed AI buildout and the reality that unilateral American restraint doesn't stop Beijing's labs from racing ahead.

This isn't purely a left-right split. PBS flagged that Trump ally Steve Bannon has said Congress needs to "rise above partisanship" because AI represents a "hinge in history." Some voices on the populist right share the underlying unease even if they reject Sanders' 20-years-in-prison, new-federal-agency approach.

The Sanders-Casar bill faces what AP bluntly called "long odds in the Republican-controlled Congress," where lawmakers have struggled to agree on far more modest AI rules. A new cabinet-level regulatory agency with criminal penalties and pre-deployment approval requirements is a massive expansion of federal power over a fast-moving industry, and Republicans controlling both chambers show no sign of taking it up.

The open question is whether Thursday's Trump-Xi meeting produces anything beyond the notification-system agreement Bessent and He Lifeng already struck. Van Hollen's letter asks for something bigger: a mutual U.S.-China pause on frontier AI development. Nothing announced so far commits either government to that.