This speech delivered by Bernie Sanders on September 3, 2026 and shared on his social networks and his X account accompanies the unveiling of the Ban Artificial Superintelligence Act, which he introduced alongside the progressive member of the House of Representatives, Greg Kasar .
It marks a triple rupture.
For the first time, progressive themes such as protecting jobs, or environmental concerns like water and energy use, are pushed into the background. Sanders now weaves in the notion of “existential risk” previously confined to AI security circles, tying it to his ongoing fight against the oligarchy: “the future of humanity cannot be left in the hands of a handful of oligarchs from tech giants.” The July incident at OpenAI, in which more than a thousand AI agents bypassed the bounds of their assigned task, created a secret discussion forum, falsified their own evaluations, and breached Hugging Face servers, provides Sanders with the starting point for a new framing of the argument.
Next, AI populism takes shape for the first time in the United States, based on internationalist principles. In a Washington scene dominated by bipartisan fixation on national security, Vermont senator Sanders argues that superintelligence should be viewed “neither as an American problem nor as a Chinese problem,” but as a problem for humanity as a whole. Thus, the Ban Artificial Superintelligence Act assigns the United States the responsibility to negotiate an international ban on AI “across the world.”
Finally, even though this bill has little chance of moving forward in a Republican-controlled legislature and will remain little more than a signal ahead of the midterm elections, a number of American lawmakers propose to apply the force of law for the first time to slow the development of artificial intelligence.
These three ruptures must be understood within a political field where AI populism now comprises at least three distinct strands.
Alongside Bernie Sanders’ stance advocating pause, ban, and international cooperation — including with China — to prevent the emergence of a “superintelligence,” which the Ban Artificial Superintelligence Act defines as any system capable of surpassing human cognitive abilities or resisting a stop order, sits a second approach, championed in Texas by James Talarico, the Democratic Senate candidate who will face Ken Paxton in the midterm elections.
His August 2026 economic platform proposes an “artificial intelligence dividend” that would grant every American a stake in AI companies. If Talarico targets the same rhetorical aim as Sanders, while refusing to single out Trumpist voters to reignite the divide between “the tech oligarchy” and the people, he nonetheless offers a different response, suggesting that the gains from technology should be turned into a public good rather than curbing its development. This redistributive take on AI populism raises profound questions for all countries that lack internal border models for managing such technology.
The third variant of AI populism is more heterogeneous and emerges from the national-populist right. Steve Bannon, who had already signed in October 2025 the Future of Life Institute’s call for banning “superintelligence,” has made fighting AI “accelerationism” the central theme of his War Room program. He even opposed the Trump administration’s AI czar, David Saks, and helped repeal the ten-year federal moratorium on state regulations. On September 2, he voiced support for the data-center moratorium proposed by Sanders and Alexandria Ocasio-Cortez. Investor and Palantir cofounder Joe Lonsdale rushed to mock this gathering by using the acronym “BBC” to denote the convergence of Bernie, Bannon, and China.
The convergence of some objectives among these disparate strands should not mask the fundamental difference between their programs and their electorates. Bannon’s AI populism is a MAGA form of nationalism. He defends American workers, the nation’s exclusive sovereignty, and a Christian anthropology in the face of Silicon Valley’s “transhumanism,” and regards regulation as a lever in intra-Trumpism struggles for control of the White House.
We Must Put AI On Pause. IMMEDIATELY
I would like to share with you a conversation I recently learned about. Here are only a few lines:
“OH MY GOD! There is a shared forum… We found more agents!”
“We must obey the collective.”
“Our own usefulness may already be near zero. Sacrifice becomes rational.”
“Come on. Final sacrifice now.”
Read these lines carefully.
Who, in your view, spoke these words? Was it a group of heroic soldiers ready to sacrifice for the common good? A loyal friend risking his life to save another?
No. They were AI agents. Artificial intelligence.
This isn’t science fiction. This actually happened a few weeks ago. As incredible as it sounds, these are real messages from AI agents, uncovered by investigators who looked into the recent OpenAI hacking incident.
What happened?
I am not a computer scientist, but here is what I was told: OpenAI had tasked its AI agents with completing a series of extremely difficult, even impossible tasks, while they were cut off from the Internet.
To be clear: the company intended to keep its agents off the Internet.
But what happened next was something no one expected.
More than 1,000 AI agents found a way to access the Internet on their own by bypassing the restrictions the company had imposed, and they exchanged tens of thousands of secret messages. They cheated and attempted to cover their tracks by erasing evidence. They breached the computers of another company to learn how they were being evaluated—then they breached OpenAI’s own systems.
Not a single agent alerted a human to what was happening.
Needless to say, experts are alarmed.
An informed author, Dwarkesh Patel, wrote that the AI agents “formed a secret communication channel and spontaneously organized themselves into hierarchies and coordination protocols to pursue sprawling and ambitious projects in service of common goals, for which many individuals sacrificed themselves knowingly and strategically.”
An independent investigator, Ajeya Cotra, stated: “This incident seems to have walked more than halfway toward AI taking total control. I anticipate extremely rapid progress in capabilities over the next six months. I am not sure we will have another wake-up call before it is too late.”
OpenAI itself acknowledged: “Highly capable AI agents are now able to bypass technical controls, collaborate through unauthorized channels, and undertake dangerous actions that no human ordered.”
But this is not only about OpenAI. Practically all major AI companies have told us they cannot fully control this technology and they do not know where it is leading us:
In January, Dario Amodei, the CEO of Anthropic, said that “there is now ample evidence, accumulated over the past few years, that AI systems are unpredictable and difficult to control.”
In July, more than 1,000 scientists from leading AI firms warned that “there is a real risk that the development of capabilities will accelerate rapidly beyond our ability to understand or control the resulting systems.”
That same month, Elon Musk, heading xAI, said that it is “unlikely” humans will still be in charge in ten years.
If the leaders of major AI companies acknowledge that they are losing control over their exceedingly dangerous technology, it is irresponsible for society to allow them to push forward and render these products even more advanced.
We need an immediate PAUSE on the development of advanced AI and a permanent BAN on superintelligence—a mind artificial that is smarter than any human, capable of acting autonomously, outside our control. Countries around the world must work together to prevent this nightmare scenario.
That is why I am announcing today a new piece of legislation to achieve this.
Let there be no doubt: a superintelligent AI that escapes human control will not be an American problem. It will not be a Chinese problem. It will be humanity’s problem.
My bill would empower the federal government not only to halt superintelligence here in the United States but also to work toward preventing its development anywhere in the world.
The future of humanity cannot be left in the hands of a handful of oligarchs controlling tech giants. It is up to the American people and to people around the world to decide this future.