Summary

  • Approximately 200 individuals demonstrated outside the offices of OpenAI, Anthropic, and Google DeepMind.
  • Activists reiterated their calls for a moratorium on the development of advanced AI models.
  • Concerns raised included AI safety, potential job displacement, energy consumption, and housing issues.

In San Francisco, protesters once again marched between the offices of OpenAI, Anthropic, and Google DeepMind, advocating for a suspension of work on more advanced AI technologies.

The protest, held on Saturday and organized by Stop the AI Race, expanded its focus beyond just AI safety to encompass worries about job losses, the environmental footprint of AI, escalating housing prices in San Francisco, and the increasing power of large tech firms.

Led by former AI researcher Michaël Trazzi, the demonstration urged major AI companies to halt the training of new frontier AI systems while keeping current technologies operational. Protesters called for a shift in research emphasis towards AI safety and alignment until more robust protections are established, with some participants advocating for enhanced regulations at both local and state levels.

Trazzi noted that the objectives of the group have evolved since their initial protest in March.

"Earlier this year, I was more focused on persuading CEOs than raising political awareness," Trazzi shared with Decrypt. "I’ve come to realize that CEOs are actually paying attention, especially after multiple blog posts and some discussions I’ve had with one of them. I also believe protests serve as a valuable tool to demonstrate public interest."

This demonstration follows a previous event in March organized by Stop the AI Race, where around 200 people marched between the locations of Anthropic, OpenAI, and xAI, advocating for a unified pause on frontier AI development. The group has maintained its momentum through ongoing protests and advocacy efforts.

Trazzi expressed optimism regarding the support the movement has garnered since then.

"I was pleasantly surprised by the endorsement from the NUHW [National Union of Healthcare Workers], and their promotion of our event on social media," he remarked. "I was also impressed by how quickly other organizations in the Bay Area, such as AI Action, mobilized to collaborate with us. I’m particularly thankful to QuitGPT for their assistance in organizing this event."

As of now, OpenAI, Anthropic, and Google DeepMind have not responded to inquiries from Decrypt.

The organizers intend to persist in their advocacy for a global pause on frontier AI advancements while pushing lawmakers to implement stricter oversight of advanced AI technologies.

This protest highlights the growing concerns surrounding AI safety.

In May, OpenAI launched enhanced safety features for ChatGPT aimed at better recognizing signs of self-harm and violence during interactions, amidst lawsuits and investigations regarding its chatbot’s handling of sensitive situations.

In June, the Trump administration mandated that Anthropic suspend access to its Claude Fable 5 and Claude Mythos 5 models due to potential cybersecurity threats. Recently, the United Nations' inaugural independent scientific panel on AI determined that scientists cannot dismiss the possibility of "catastrophic harm" as AI technology evolves more rapidly than scientific comprehension and regulatory frameworks.