Protesters Hit OpenAI, Anthropic, xAI in AI Safety March

Stop the AI Race brought roughly 200 protesters to San Francisco's AI corridor on Saturday, marching past the offices of Anthropic, OpenAI, and xAI in what organizers called a push to force the industry into a collective conversation it has so far managed to avoid. The demonstration drew researchers, academics, and members of advocacy groups including the Machine Intelligence Research Institute, PauseAI, QuitGPT, StopAI, and Evitable — a coalition that, at first glance, looks like a fringe movement but increasingly reflects a current of anxiety running through the field itself.

What Is Stop the AI Race Asking For?

The core demand is a conditional pause — not a permanent shutdown, but a coordinated stop to building increasingly powerful frontier models until other major labs credibly commit to the same. Michael Trazzi, the Stop the AI Race founder and documentary filmmaker who organized Saturday's march, put it plainly: if the U.S. and China agreed to stop racing toward more dangerous models, both countries could redirect that energy toward safety-focused work instead — medical AI, alignment research, interpretability.

The march kicked off at noon outside Anthropic's San Francisco offices, then wound past OpenAI's headquarters and finally to xAI. At each stop, speakers from the participating organizations addressed the crowd. The rotating-stage format was deliberate — Trazzi wanted each group to have a public moment rather than one organization dominating the message.

"There are a lot of people who care about this risk from advanced AI systems," Trazzi said. "Having everyone marching together shows people are not isolated in thinking about this by themselves."

Because we're in this race between companies and countries to build the systems as fast as possible, we're taking shortcuts and cutting corners on safety. There is never a race that has no winners. What we have is a system we cannot control, and that's why it's called a suicide race.

The Protest Arrives at an Awkward Moment for the White House

The timing is not incidental. Just days before the San Francisco march, the Trump Administration published its Trump national AI legislative framework — a national standard for laws governing AI development that the White House explicitly framed as a commitment to "winning the AI race." That language is the exact opposite of what Saturday's protesters were calling for. The administration's argument — shared by many in Washington — is that slowing U.S. AI research hands an advantage to China. Trazzi's counter: nobody, including China, actually wants systems they cannot control.

It's a tension that isn't going anywhere. And frankly, the protesters may be making a more coherent argument than they're given credit for. The "we can't pause or China wins" framing assumes the race ends somewhere — that there's a finish line where the winner gets a trophy and the risk disappears. Trazzi's framing assumes no such thing.

The history here matters. In March 2023, the Future of Life Institute published the Pause Giant AI Experiments open letter calling for a moratorium on AI enhancements beyond what GPT-4 could do. The signatories included xAI founder Elon Musk, Apple co-founder Steve Wozniak, and Ripple co-founder Chris Larsen. That letter now has over 33,000 signatures — and Musk, notably, has since launched xAI and is now one of the companies being protested outside.

How Would You Even Verify a Pause?

This is the question that tends to sink pause proposals in policy circles — verification. Trazzi has a concrete answer: limit compute. If regulators cap how much computing power any lab can use to train new models, they effectively cap the frontier. It's not airtight, but it's more actionable than most of the alternatives floating around Washington right now.

Saturday's protest follows a pattern of escalating direct action from this corner of the AI safety movement. In September, Trazzi staged a week-long hunger strike outside Google DeepMind's London offices. A parallel hunger strike by Guido Reichstadter ran simultaneously outside Anthropic's San Francisco location. These are not casual observers — they're people willing to put their bodies on the line over timelines they believe are genuinely dangerous.

Trazzi said further demonstrations are planned in cities where major AI companies operate. The stated goal isn't just to get press coverage — it's to reach employees directly. "We want to show up where the employees are," he said. "We want to talk to them, and we want them to talk to their leadership." The bet is that whistleblowers and conscience-driven engineers have leverage their employers don't want to acknowledge. OpenAI, Anthropic, and xAI did not respond to requests for comment.