OpenAI, Anthropic, Google Draw Anti-AI March in San Francisco

More than a hundred people walking between the front doors of OpenAI, Anthropic, and Google is not, in absolute terms, a large protest. But it is a new kind of scene for San Francisco, and the parade-style anti-AI march covered by The San Francisco Standard is worth reading as a marker rather than a headcount.

The roughly two-hour route started at OpenAI's Mission Bay headquarters, threaded through Anthropic's downtown office, detoured past Andreessen Horowitz on Townsend Street, and ended at Google's San Francisco office on the Embarcadero. Organizers called it 'Freeze AI on Slushy Day,' riffing on 7-Eleven's annual free-slurpee promotion. There was a brass band, and a Barney the dinosaur costume worn by Rohan Prasad, 26, who told the Standard he had pivoted from AI safety research to activism and picked the costume as a commentary on not wanting humans to go extinct like the dinosaurs. Signs ranged from 'Stop the race, it's Paw-sible' to 'Pls Do Not Kill Me,' and San Francisco police escorted the march on motorcycles, in SUVs, and on foot.

Organizer Michael Trazzi, 30, framed the concern as 'building an AI that is so smart that we can't control it.' Trazzi is the same person who conducted a hunger strike in September in front of Google DeepMind's London office, so the theatrical tactics are being ported across cities by a small but persistent core rather than a mass movement. The coalition groups named in the reporting include Stop The AI Race and QuitGPT.

For the labs, the honest signal is not the crowd size but the venue selection: OpenAI, Anthropic, Google and Andreessen Horowitz on a single two-hour walk, with all three of the labs declining to comment. What the reporting does not give you is a durable-membership count for the coalition, any funding trail behind the coordinated bus from Berkeley, or a read on whether current lab employees are joining. Those are what would matter for whether this scales or fades.

The forward-looking piece is that 'AI safety' is walking off the whiteboard and onto the sidewalk. When former safety researchers put on Barney costumes and Erik Leklem, 52, brings his two young children from Oakland because he says he is worried about their future with this technology, the labs' comms and policy teams have a different job than the one they have been staffing for. Whether they respond by engaging or by hardening the perimeter is the more interesting question for the rest of the year.