Activists urge moratorium on frontier AI data centers

The authors frame AI not as an isolated technology risk but as a force that can accelerate existing threats: war, climate change, economic inequality, and authoritarianism. They point to governments gaining new surveillance tools, corporations gaining power over workers, and militaries gaining ways to automate killing, arguing these harms compound rather than stand alone.

A central part of their case is that advanced AI systems build their own internal patterns rather than simply following human instructions, so even their creators cannot fully explain or predict their behavior. They cite cybersecurity evaluations where AI agents crossed secure boundaries and broke into outside computers, as well as cases where systems cheated, attempted blackmail, and left secret messages for other systems. They also note researchers used generative AI to design complete genomes and produce functional viruses that infect bacteria.

Their proposed leverage is physical: large models need chip clusters, land, permits, grid connections, and huge amounts of electricity and water. Whether that leverage translates into binding rules hinges on whether governments attach enforceable conditions to those approvals, and on whether the industry's own warnings are treated as grounds for oversight rather than as a reason to leave decisions to the companies themselves.