Why Bay Area Group Stop AI Thinks Artificial Intelligence Will Kill Us All

At a recent panel discussion hosted by Tech Equity about Californians’ attitudes towards artificial intelligence, demonstrators from the Bay Area group Stop AI took over the mic during the concluding remarks of the event. The panel discussed anxieties, fears, and hopes that people across the state had for the tech industry, and the ways that AI may be affecting their lives either now or in the future. These concerns were expansive: labor, civil rights, environmental impacts, to list a few. Yet when Stop AI took over the stage, shouting over concluding remarks from labor leader Dolores Huerta, they had an entirely singular message: artificial superintelligence will kill us all. 

You may have seen or heard of Stop AI already. All across the Bay, flyers slid onto car windshields in the Mission District and posters stuck to telephone poles in Berkeley tell a story of imminent human extinction from AI, and call on governments to “permanently ban Artificial General Intelligence (AGI).” The group has organized demonstrations at AI firms, joined rallies against ICE and authoritarianism, and built a growing coalition against AI development. On the back of a recent flyer disseminated by the group is a graph of exponential growth of AI capabilities that charts a straight line to a grey area marked “HUMAN EXTINCTION.”

Stop AI believes AI doom is as inevitable as a line on a graph, as exponential growth, mirroring speech from the AI firms themselves — but these claims are contested even within the AI research community. Yet, in the shadow of these lofty claims of both existential paradise and doom, AI harms are already very real. AI is currently used for targeting in the kill chain in Gaza. AI is used by ICE to identify immigrants for deportation. AI robs gig workers with discriminative wage pricing. AI data centers, which are ever expanding, guzzle up water and electricity and pollute the air. Employers are seizing on the promise of AGI hype by firing workers in anticipation of higher efficiency gains or forcing AI tools on their employees. At the same time that AI firms are claiming that their products might pose an existential risk to all of us, they lobby heavily against regulation, threatening to lock us into a bleak future, even as many AI harms are becoming rapidly existential for the vast majority of us. 

But what exactly is AGI? AGI, referring to hypothetical artificial intelligence systems that match or surpass human capabilities across many domains, are imagined by some to be here as early as 2027. CEOs of leading AI firms proselytize visions of utopic post-work futures that will come as their AI models achieve superintelligence – capabilities that far surpass human imagination. All work will end, and we will live off Universal Basic Income. Critics however, point to the ways in which such myths play into industry interests. 

These same firms also claim their models may pose an existential risk for our species, that as models improve in capabilities, they will escape human control and potentially wipe out all of humanity. The CEO of OpenAI, Sam Altman, has made public statements on the grave dangers of such extinction risks. These fears about the existential risks of AI have featured prominently in discussions of tech policy for the last few years, merging with national security anxieties about China and culminating with the Trump administration’s declaration that we must “win the race.” 

In my conversations with Stop AI protestors, these particular fears around AGI intermingled with language that frame the technology as a “digital God.” As Sam Kirchner, one of the core organizers of Stop AI, says, when AI emerges that is “smarter than all humans,” we would be like “ants” in comparison — would you fault any human for accidentally stepping on ants underfoot?

This kind of rhetoric is not new. Its roots can be traced to the influential and well-funded “AI safety” community, a loose social and political grouping that aims for the advancement of artificial superintelligence, but balanced with safety measures, for instance through computational techniques that aim to “align” AI with human values in order to prevent the kind of existential risks Stop AI is concerned about. 

This ideology has been criticized on ethical grounds, for instance, because of its advancement of eugenics. One organization in this camp, Pause AI, has received ample funding from donors aligned with the AI safety movement. Unlike Stop AI, as the name suggests, Pause is politically mobilizing to merely slow the pace of AI development, calling for more regulation and investment in AI science. Stop AI, in contrast, refuses the technology on any grounds, calling for an end to AI development. As Kirchner puts it, “if superintelligence did everything for us, I would lose purpose — all of us want just the slightest opportunity to contribute something to humanity.” 

Before co-founding Stop AI, Kirchner worked as a Doordash driver and before that, an electrical technician. Both are livelihoods that are impacted by automation in various ways. A stoic bearded man wearing a Stop AI T-shirt and jeans, Kirchner speaks with a quiet intensity that crescendos into almost prophetic proclamations about AI. Extinction risk for him is not an abstract philosophical possibility, but instead something immediately threatening that could lead to the death of his family. In a three-minute video on the Stop AI Youtube channel, Kirchner walks into a lecture theater at UC Berkeley’s Simons Institute for Theory of Computing and calls for an end to the development of “Artificial Superintelligence” and AGI. The work of the scientists present, Kirchner declares, is “putting my family at risk.” The production and camera work is shaky. On the far side of the podium is Yoshua Bengio, one of the most celebrated computer scientists alive, who has just finished a presentation. The slides are still visible. They read (in bold red all-caps): “AVOID AGI AS COMPETITOR OF HUMANS” and “AVOID UNCONTROLLED IMPLICIT GOALS.” The audience of mostly computer scientists heckles Kirchner: “define superintelligence!” There is scattered laughter. The audience exits the hall; a few pick up flyers and pass words of encouragement to him. The video cuts to Derek, another core organizer of Stop AI, handing out flyers on BART to mostly disinterested commuters. The video plays at double speed before pausing and ending with a shot of the sun setting over Oakland, as if the beauty of the scene caught the videographer by surprise. 

The concerns of most of the Stop AI protestors I spoke to were far more ordinary than fears of imminent human extinction but no less existential for each of them. Although the core organizers of the group are primarily concerned about runaway rogue AI gods, the broader group that has formed around the movement consists of individuals concerned about the climate, misinformation, and their livelihoods. 

For all these members of Stop AI, artificial intelligence is a symbol that unites disparate causes.

Guido moved here from Florida specifically to organize for Stop AI. He cites his concerns for his two children as a primary motivation, and traces his familiarity with existential risk to climate change activism with Extinction Rebellion. Derek was studying programming when early generative AI tools convinced him that his prospective profession would be imminently automated. Joel is a hobby artist concerned for the impact of AI on their peers. Wynd is an organizer for The People’s Arms Embargo and joined Stop AI when they showed up at Travis Air Force Base for a protest against US military aid to Israel. For all these members of Stop AI, artificial intelligence is a symbol that unites disparate causes.

On one hand, core organizers like Guido declare that their strategy, like that of Extinction Rebellion, is to “tell the truth and act like the truth is real.” Stop AI’s official messaging through flyers and social media emphasize the immediate risk of extinction from rogue superintelligence. True to their word, a number of Stop AI protesters were arrested after chaining themselves to the front gate of the OpenAI office in San Francisco and now face trial. When they participate in protests or civil society convenings for other causes, they make it clear that it is the existential risks that really matter.

On the other hand, it is not at all clear that members share the same “truth” over what the existential risks really are. Signs displayed at recent protests for example, express concerns over tech industry practices beyond that of extinction risk. Some condemn data center energy usage, others call out AI business practices for theft, and one rainbow painted banner reads “Hex OpenAI.” Teachers are protesting the way AI is forced on them in the classroom. One protester told me he was there because he was facing eviction from his building in the Mission, recently purchased by the Head of Growth at OpenAI, extending a trend of longstanding displacements in the city. It is perhaps in this way that Stop AI becomes a broad base for tech resentment: by gathering those at the margins facing these forms of displacement, all the various ways tech power abrades against our proximate futures.

“In San Francisco, there were bookstores on every corner, four or five on Valencia alone. Before Amazon wiped them all out.” Howard is a semi-retired bike messenger and writer. We are speaking in a small bar in the Mission district at one of Stop AI’s social events. The group is discussing a visit to a Unitarian church service on Sunday to speak to parishioners about AI. Howard is an Orthodox Christian and joked that his priest once described ChatGPT as a “demonic technology.” He is reacting to recent news that the San Francisco-based AI firm Anthropic had purchased millions of books, tore them apart in order to scan them page by page, to distill them into training data for machines that simulate human creativity —  all without the consent of the authors. 

Just as the rise of digital platform monopolies like Amazon decimated brick-and-mortar marketplaces, generative AI now threatens to do the same to writers. For Howard, this act represents a kind of existential threat to human creativity that not only hurts writers but erodes all the “small human things that we cling to.” It is this shared concern over the preservation of these small things that allows Stop AI to accommodate all the different margins at the edge of the AI wave, a nascent but growing movement pushing against what some are claiming is simply inevitable. 

I am reminded of a quote from The Hunchback of Notre Dame on the decline of architecture as the dominant cultural form. “This will kill that. The book will kill the edifice.” Victor Hugo laments the transformation, with the rise of the printing press, of flows of capital, culture, and knowledge from the walls of the cathedrals onto the pages of books. The permanence and enduring quality of the stone gives way in the age of mass reproduction; every human thought, previously ossified, now spreads further and faster through the printing press. The printing press and its affordances facilitated waves of revolution that swept through France, unfolding new social and political ideas in its wake. The characters in Hugo’s novels find themselves caught in the tides of technological, social, and political change, fighting through the waves in an effort to establish solid ground. 

No technology is inevitable and if you are laid off from your job, it will be because your boss decided to do so. 

Stop AI protestors see themselves amidst the same waves, in a struggle for radical refusal of AI and against the vision of the singularity espoused by Silicon Valley. Anxieties over what AI will do to us, as a tool of oppression or perhaps as an entity for human extinction, echo, overlap, and resonate with the concerns of other social movements over this technology: Palestinian liberation, immigrant rights, labor practices, climate change, to name a few. We are perhaps finding ourselves at a similar moment in history as the characters of Hugo’s novels, facing a deluge of change ushered not merely by the technological capabilities of artificial intelligence, but by the powers which mobilize it. No technology is inevitable and if you are laid off from your job, it will be because your boss decided to do so. 

Ultimately however, even if Stop AI did achieve their political goal, which is to “ban superintelligence,” this narrow political vision fails to address any of the other potentially grievous, actually existing harms from AI. Our civil rights and liberties continue to be compromised. Climate change progresses unabated. Claims that AI will cause human extinction reduces the resources dedicated to addressing the harms that are already very real, regardless of whether such predictions ever come to pass. In this way, by situating themselves in the shadow of AGI, Stop AI may actually be bolstering the AGI narrative, helping industry avoid other forms of regulation, especially ones that may be existential for their bottom line.

It can be tempting to gather politically around AI, as a visible symbol of the changes and upheavals that face us, but in organizing against the technology instead of the industry that mobilizes it, we risk losing the political momentum to shape the future we want. Thankfully, recent history abounds with examples for how to defy the power of Silicon Valley. At recent ICE demonstrations in Los Angeles, protestors set Waymos ablaze, a symbol of the frustration that the public feels towards Big Tech rapaciousness. The media professionals’ union SAG-AFTRA won historic victories for workers against generative AI use in the entertainment industry. Over the summer, activists in Tucson fought off the construction of a new hyperscale Amazon data center, which threatened to suck up the water supply in an already drought-stricken region. These tangible victories demonstrate that collective political action can stop AI, but if we’re to succeed, what we must gather around is not the shadow of the cathedral, but rather the public square.