3 police officers arrest white man in brown shirt at AIdemonstration
Stop AI co-founder Guido Reichstadter was one of four people arrested outside of Anthropic’s headquarters on Thursday afternoon. Photo by Zoe Malen.

As the markets roared over Anthropic’s IPO debut Thursday — minting a whole new crop of paper millionaires in San Francisco — more than 100 people demonstrated outside the AI giant’s Howard Street headquarters to remind the public that the company had invoked the end of the world to get to this lucrative moment. 

By 2:30 p.m., Mission Local saw at least four of the protestors in handcuffs for blocking the front doors to the building lobby. As one was dragged away, the crowd, which had vastly thinned out by the afternoon, chanted “Stop AI!” 

After the arrests, police officers formed a wall around the front door to the headquarters, while demonstrators with the organization Stop AI continued to hold signs and pace around the building. SFPD Lt. Mark Moreno said the four arrested demonstrators — one of them Stop AI co-founder Guido Reichstadter — will be given a ticket and court date and then released.

A technology with the capability to unleash disasters upon humanity was actually part of the selling point in the company’s paperwork to go public last month. As Anthropic sought to justify a trillion-dollar valuation and its costs, the company warned that advanced AI systems could act on their own and — without proper guardrails — “pose catastrophic or existential risks to humanity.”

The effects of heavy rains this week kept the air brisk outside Anthropic’s office building, despite a parting of the clouds that offered a brief respite during the demonstration. Protesters largely outfitted in rain gear called for a ban on Recursive Self Improvement, or RSI, a process in which AI systems design, train, or refine themselves, and accelerate progress toward uncontrollable artificial superintelligence (ASI).

Demonstrators — loaded with  dire signs — chanted toward the glass walls of the office at 500 Howard Street: 

“Ban RSI, We don’t wanna die.”

“ASI is not a race, build it at a human pace.”

A crowd of protesters holding signs, including a yellow diamond sign that reads “Slow Humans at Play” and a large blue sign reading “Freeze Now.”
Demonstrators outside Anthropic headquarters on November 19. Photo by Zoe Malen.

Office security had  attempted to establish a trespass perimeter outside the building, marking the tiled sidewalk with black tape, and demonstrators for the most part stayed outside those lines. 

But by 12:30 p.m., close to a dozen people crossed the lines and sat down against the glass windows of the office’s front door. These protesters were markedly older than the crowd of 20-and 30-year-olds  who had picked up signs and watched from behind the tape. Among those who breached the perimeter was Wynd Kaufmyn, a 69-year-old demonstrator with StopAI who became the first AI protester to be jailed in San Francisco this year, after she and other members chained and locked the front doors of OpenAI’s headquarters. 

“We plan to stay here until they stop,” Kaufmyn said, gesturing to the Anthropic offices behind her. Kaufmyn shrugged at the possibility that  she and other protesters might be forcibly removed or arrested: “It wouldn’t be my first time.”

Office security urged the sitting demonstrators to leave, and tried to contain the crowd of bystanders and reporters that swarmed them. Inside the front lobby, people watched from the other side of the glass. 

San Francisco Police Sgt. Bryan Kneuker knelt and one by one urged the demonstrators to leave. He told a reporter that Anthropic security had said the company intended to  go after  people who refused to remove their bodies from the front-door windows.

“What these folks are doing — and if they’re arrested for a crime — it would pale to the crime these AI companies are committing by endangering our lives,” said David Kruger, an AI safety advocate who looked on from behind the tape before walking up for a closer look. Kruger is the founder of Evitable, a nonprofit organization that advocates for a moratorium on the development of AI.

By 1:39 p.m., supporters  had brought pizzas for the demonstrators sitting against the windows. By 2:30  p.m., four of the protesters were in handcuffs.

Speakers at the demonstration, including leading AI safety advocate and former OpenAI researcher Daniel Kokotajlo, said they believe AI firms could be half a year away from achieving RSI. 

“Probably six months. Maybe it’s 12 months. But whether they actually do it is a separate matter, because our goal is to stop them — to achieve an RSI ban,” Kokotajlo told Mission Local. “It does feel like there are a bunch of people in the industry who are starting to get concerned.”

Former Anthropic researcher Jacob Coxon, whose public statements about AI firms gambling with people’s lives helped highlight AI safety issues, said he was confident his former coworkers felt the same.

The people still inside Anthropic “don’ t want RSI to happen,” Coxon told Mission Local .

Security guards stand at a glass entrance, one using a phone to photograph people sitting on the floor behind the door.
Security officers and demonstrators at the front doors to Anthropic headquarters on November 19. Photo by Zoe Malen.

Brandon Pho covers tech and power in San Francisco.

He has reported on communities up and down California since 2018.

Right before joining Mission Local, he sparked a California lawsuit against the federal government after obtaining and publishing the blueprints for a secret Bay Area ICE facility.

He likes the TV series ‘Columbo’ and lives under the panoptic surveillance of his two cats in the Mission.

Leave a comment

Please keep your comments short and civil. Do not leave multiple comments under multiple names on one article. We will zap comments that fail to adhere to these short and easy-to-follow rules.

Your email address will not be published. Required fields are marked *