Residents in San Francisco Protest Against the Frontier AI Race Citing Safety Concerns

Human Rights Research Center
10 hours ago
3 min read

Author: Amy Asubonteng

March 30, 2026

HRRC echoes the concerns of protestors in San Francisco and emphasizes that AI data centers make up a significant amount of global carbon dioxide emissions and contribute to environmental degradation. Continuing the development of frontier AI poses an even greater risk to international environmental and human rights.

Protestors in San Francisco, California, marching from Anthropic to OpenAI to xAI, demanding a collective agreement on the pause. [Image credit: AP Photo/Michaël Trazzi]

On Saturday, March 21, dozens of protestors marched together outside the headquarters of OpenAI, Anthropic, and xAI to call for a pause on the development of frontier AI systems. The protesters demanded AI labs collectively agree to a pause under the condition that all AI-focused corporations around the world do the same.

Organized by Stop the AI Race, led by former AI researcher and filmmaker Michaël Trazzi, the protests raise the alarm that further development of AI— especially creating AI that is self-sufficient and capable of automating research— poses significant risks, including the possibility for AI systems to transcend human control, negative effects on employment, and safety risks.

Trazzi further emphasized his belief that building AI capable of automating AI research and self-improvement can be a danger to the human race, even risking human extinction. “It's not only me and other researchers saying this, it's the lab CEOs themselves that the risk is real,” He said.

Collective action to slow the development of frontier AI remains challenging. Companies are reluctant to limit the speed of development unless it is clear that all other labs, including Chinese companies, halt development as well. ’The race has a political dimension, too: the National Policy Framework for AI back issues in December 2025, Ensuring a National Policy Framework for Artificial Intelligence, proposes global AI dominance for the U.S., and encourages companies like OpenAI, Anthropic, and xAI to produce innovative AI, especially to compete with China.

However, some AI companies are making some steps to address AI risk. In February, Anthropic released its original Responsible Scaling Policy (RSP), a framework to manage the risk of AI, including by refusing to train models more powerful than its current model unless safety measures were guaranteed. However, it was replaced with another RSP (v3.0) that same month, which represented a more non-binding commitment. This move was likely made because pauses in AI development by one company will not limit AI development if competitors remain involved in the race. For this reason, protesters have demanded that all competitors cease their work on advancing AI.

Glossary

AI (Artificial Intelligence): the branch of computer science involved with the design of computers, robots, programmed devices, and software applications having the capacity to imitate human intelligence and thought.
AI Race: the competition between countries and corporations to become the first to create and own the most advanced AI systems.
Data Centers: a facility, or building, containing servers meant to store, process, and distribute information, or data, across the internet.
Carbon Dioxide: a heavy colorless gas linked to global warming that is formed in combustion, in the respiration of animals, and employed by plants.
Framework: a structure, openwork, or frame used to systemize concepts, ideas, rules, or beliefs.
Frontier AI: AI models that are the most advanced at a given moment, outperforming existing cutting-edge systems.
Non-binding commitment: a commitment, or an agreement, to which the parties (or party) involved are not obligated to follow.
Protest: the act of objecting, usually an organized public demonstration of disapproval
Self-sufficient: the ability to maintain oneself or itself without outside help.
Self-improvement: can refer to a computer system's ability to enhance its own intelligence and capabilities without outside help.
Transcend: to overpower or go beyond the limits.

Sources

HRRC accepts NO funding from governments for our research or campaigns. At every step of the way, our impact is made possible because of people like YOU. Make your contribution now!

Residents in San Francisco Protest Against the Frontier AI Race Citing Safety Concerns

Recent Posts