Nonprofits sue OpenAI and Anthropic to halt development of more powerful AI

Four U.S. nonprofit groups sued the organizations behind ChatGPT, Claude, Gemini, Meta’s AI models and SpaceX’s AI business on November 12. They asked a California state court in San Francisco to immediately bar the defendants from developing or deploying AI systems more capable than the strongest systems each had on the filing date. Public Citizen, one of the plaintiffs, announced the lawsuit that day. The plaintiffs’ lawyers published a signed copy of the complaint.

The filing has produced no confirmed change for people using ChatGPT, Claude or Gemini. The first requested injunction sets a baseline for each defendant’s AI capabilities at the time of the complaint. Its wording specifies neither a product version nor a test for deciding whether a later system exceeds that baseline. The baseline is each company’s technical capability on the filing date, independent of the names under which its products are sold.

Public Citizen brought the case with Irreplaceable Education Fund, Heat Initiative and the Alliance for Secure AI, which appears on the complaint under the legal name Future Tech Insights. The case names seven entities across five corporate groups. It lists OpenAI Group and OpenAI Foundation separately, as well as both SpaceX Super Intelligence LLC and SpaceX Super Intelligence Corp. The public copy is signed and dated November 12. Its case-number field is blank.

A separate request targets how AI research is done

The plaintiffs also seek an immediate injunction barring the defendants from using AI to independently carry out or direct research and development of more capable systems. That request concerns the part AI plays in producing a future system. The first request concerns the capability of the system being developed or deployed. The proposed wording leaves open how particular research tasks would be classified.

The complaint asks for additional orders against practices that the plaintiffs say create a public nuisance or violate California’s Unfair Competition Law. It also requests attorneys’ fees. The groups cite publicly reported AI security incidents in arguing for an order before the companies develop or deploy more capable systems. Their complaint sets out allegations about the companies’ conduct and the risks that they say it poses.

The legal basis in California

California’s Unfair Competition Law covers unlawful, unfair or fraudulent business practices. Section 17203 of the state’s Business and Professions Code gives courts authority to stop conduct that falls under the law. Section 17204 requires a private plaintiff to show an actual injury and loss of money or property caused by the unfair competition. The complaint links the challenged conduct to alleged threats to public health and interference with the use of property.

The groups say responding to the risks drew resources away from other work. Public Citizen describes reassigning staff time to evaluate policy and legislation. Heat Initiative, which works on child safety, says it redirected resources from other campaigns to monitor frontier AI development and coordinate with partner organizations. The plaintiffs cite those costs in support of their right to bring the case.

The complaint also cites California’s existing Transparency in Frontier Artificial Intelligence Act as evidence of state policy on safe development. The state attorney general says that law requires covered large developers to put a safety framework into practice and publish it, including how they assess and reduce severe risks. The groups seek the injunctions under the Unfair Competition Law.

California Courts describes filing and serving a civil complaint as steps that give the defendants a chance to respond. The complaint and the plaintiffs’ announcement disclosed no order granting a freeze or date for a hearing. A Meta spokesperson declined to comment to Reuters on November 12. The outlet said the other defendants did not immediately respond.