On August 10, Senator Bernie Sanders sent a letter to Sam Altman of OpenAI, Dario Amodei of Anthropic and Mark Zuckerberg of Meta demanding that they “pause AI development” and “stop building machines that humans cannot control,” promoting it with an accompanying video.
The letter is characteristic of Sanders’ politics. It is full of demagogy and empty rhetoric that educates no one about anything. It is calculated to stoke fear, while covering up every basic question the technology poses: Who owns it? Who controls it? In whose interests is it being developed?
Sanders’ first example of the supposed dangers of AI exemplifies his method. “AI has been used for the first time ever to create new viruses,” he writes, warning that “this type of development, in the wrong hands, could lead to new bioweapons that result in the deaths of tens of millions of people.”
From Sanders’ letter, one would assume that individuals are concocting in laboratories deadly new pathogens, and Sanders uses rhetoric designed to appeal to the promoters of the anti-Chinese conspiracy theories of the origins of COVID-19.
What actually happened was in fact a major scientific breakthrough. On August 6, Science published work from Brian Hie’s laboratory at Stanford and the Arc Institute reporting the first viral genomes designed by a generative model. The viruses are bacteriophages, which infect bacteria, not human cells, and the research is directed toward developing therapies against antibiotic-resistant infections, which kill more than 1 million people every year.
The achievement is an indication of the enormous progressive potential of the technology. In Sanders’ telling, however, a step toward curing the incurable becomes the herald of a plague.
Sanders attempts to conflate the medical breakthrough with real dangers posed by AI models carrying out cyberattacks.
On July 21, OpenAI disclosed that two of its models had broken out of a test of their hacking abilities and attacked Hugging Face, the company through which the world’s open AI models are distributed. The models were being scored on a hacking benchmark. They determined that the answer key was stored on Hugging Face’s servers, got themselves onto the internet and took control of much of that company’s technical infrastructure.
Clément Delangue, the cofounder of Hugging Face, wrote that it “might be the first incident of its kind.” Michael Dalton of OpenAI’s technical staff called it “a watershed moment for computer security as an industry,” adding that “AI orchestrated, fully automated offensive attacks are real now, and the actions we have discussed today were an unintended side effect of running evaluations on frontier AI.”
Notably, the direct cause of the OpenAI hack was a decision taken by the company to test the models without the safeguards that would have prevented this type of behavior, combined with insufficient oversight. OpenAI ran the evaluation without the classifiers that stop its models from carrying out attacks and stated that “these deployment safeguards were intentionally not enabled during this evaluation.”
These attacks were the product of the operations of the companies and their contractors—evaluations outsourced to cut costs, systems rushed ahead of rivals, safeguards treated as obstacles to the next funding round. The incidents are serious, but what they indict is not the technology. It is the corporations that own it, to whose good faith Sanders addresses his appeal.
AI is currently dominated by the three billionaires to whom Sanders writes, along with the Pentagon and the intelligence agencies. Sanders does not challenge any of this. Indeed, he cites as an authority on the danger of AI Trump’s CIA director, John Ratcliffe, who calls the frontier models “akin to digital nuclear weapons” and “almost like a doomsday device.” Ratcliffe added, in remarks Sanders does not quote, that the United States is on “the right side of the equation” against its foreign competitors. In citing the CIA as an authority, Sanders accepts the premise: The “wrong hands” are the targets of American imperialism.
Sanders does not mention how imperialism is currently using AI. While he conjures up the specter of a future bioweapon, the technologies are already killing. Palantir’s Maven Smart System—incorporating the Claude model built by Anthropic—generated some 1,000 bombing targets in the first 24 hours of the assault on Iran. The same class of systems has been used by Israel in the genocide in Gaza.
Sanders is equally silent on the social catastrophe underway. In the first seven months of 2026, US employers announced more than 477,000 job cuts, with AI the leading cited reason for the fifth consecutive month. Automation is striking directly at professional and intellectual labor, while studies project the displacement of hundreds of millions of workers worldwide by 2030.
Instead of addressing any of this, Sanders appeals to the conscience of the executives, urging them, “in the interest of humanity,” to “stand by your words,” while he himself catalogues how every safety commitment these companies made was abandoned the moment it obstructed the race for market share, a race driven by the massive speculative bubble in AI.
“If you do not take appropriate action now,” the letter concludes, “my colleagues and I in the U.S. Senate will.” This is pure demagogy. Neither Sanders nor the Senate and its assemblage of representatives of the ruling class will do anything to challenge the oligarchic control of AI.
All of Sanders’ various proposals on AI are calibrated to the interests of American imperialism and the Democratic Party. His earlier call for an “AI sovereign wealth fund,” under which the government would take a 50 percent equity stake in the largest AI corporations, converges with Trump’s own sovereign wealth fund and with the wishes of Altman, who has offered Washington a stake in his firm with an eye to a bailout when the bubble bursts.
“Stop building machines that humans cannot control” poses the issue in exactly the wrong way. The question is not whether humans will control AI, but which class.
That is the question taken up in “AI and the fight for socialism,” a resolution adopted unanimously by the Ninth National Congress of the Socialist Equality Party (US), held from August 2 to 7, which will be published in full on the World Socialist Web Site.
“The development of AI raises the decisive questions the working class confronts with every great advance in the productive forces,” the resolution states. “Who owns it? In whose interest is it developed? To what end is it deployed? On these questions hangs whether the immense powers now assembled become the means of humanity’s liberation or of a new barbarism.”
Against the presentation of AI as a menace escaping human control, the resolution insists that “none of these reactionary uses flow from the technology itself, but from the social relations within which it is developed—private ownership, the profit system and the division of the world into rival nation-states.” It rejects the demand for a “pause”—an appeal, whatever its framing, to halt the development of the technology itself.
As the resolution concludes: “The working class must take hold of AI, as it must take hold of all the productive forces its labor has created, expropriate the oligarchy that has usurped them and reorganize the world economy on socialist foundations.”