Approximately 200 demonstrators gathered in San Francisco on Saturday, marching past the offices of three major artificial intelligence companies to demand a conditional pause on the development of increasingly powerful AI systems. The protest was organized by Stop the AI Race, a group founded by documentarian Michael Trazzi. Participants included researchers, academics, and members of organizations such as the Machine Intelligence Research Institute, PauseAI, QuitGPT, StopAI, and Evitable.
The march began at noon outside the offices of Anthropic, before moving on to OpenAI and then to xAI. At each location, speakers from the participating organizations addressed the crowd. Trazzi told reporters that seeing so many people march together was intended to show that concern about advanced AI risks is widely shared, not isolated to a small number of individuals.
The central demand of the protest is for AI companies to halt construction of new frontier models and redirect resources toward safety, provided that other major laboratories credibly commit to doing the same. Trazzi also called for international treaties between AI developers in different countries. He argued that if the United States and China agreed to stop building more dangerous models, both nations could instead focus on beneficial applications such as medical AI.
Saturday’s demonstration is the latest in a broader series of efforts to slow AI development. In March 2023, the Future of Life Institute published an open letter calling for a moratorium on further enhancements to leading AI tools, following the public launch of ChatGPT the previous year. That letter, titled “Pause Giant AI Experiments,” has since gathered more than 33,000 signatures from figures including Elon Musk, Steve Wozniak, and Chris Larsen. Trazzi himself staged a week-long hunger strike outside Google DeepMind‘s London offices in September, while a parallel hunger strike took place outside Anthropic’s San Francisco offices.
Opponents of a pause, including government officials and advocates of continued AI investment, warn that slowing research in the United States could hand an advantage to international competitors. Last week, the Trump Administration released an AI policy framework aimed at establishing a national standard for laws governing AI development, framing it as a commitment to winning the global AI race. Trazzi pushed back on that framing, arguing that the current competitive dynamic between companies and countries is leading developers to cut corners on safety, resulting in systems that no one can fully control.
Even if a pause were agreed upon, verifying compliance would present significant challenges. Trazzi proposed that limiting the computing power available to train new models could serve as a practical verification mechanism, since compute constraints would effectively restrict the development of more capable systems. He added that whistleblowers within AI companies could play an important role, given that they are directly involved in building the technology.
Following the San Francisco march, Trazzi indicated that further demonstrations are being considered in other cities where major AI companies maintain offices. The goal, he said, is to engage directly with employees and encourage them to raise concerns with their leadership from within their organizations. OpenAI, Anthropic, and xAI did not respond to requests for comment on the protest.
Originally reported by Decrypt.
