Lawsuit Targets OpenAI, Anthropic and Three Other AI Giants, Alleging They Cannot Control Their Technology

SAN FRANCISCO: A coalition of nonprofit groups on Thursday sued five major artificial intelligence companies, seeking to halt the development of advanced AI models on the grounds that the technology poses severe risks to humanity.
The suit was filed in San Francisco Superior Court by the Alliance for Secure AI, together with the Irreplaceable Education Fund, Public Citizen and the Heat Initiative. It names Anthropic, Google, Meta, OpenAI and SpaceX Superintelligence (SpaceXSI) as defendants. None of the companies immediately responded to requests for comment.
Brendan Steinhauser, CEO of the Alliance for Secure AI, accused tech chiefs of continuing to build ever more powerful AI systems even after acknowledging potentially catastrophic outcomes. He said the companies are accelerating development and commercial deployment without guarantees of safety or reliable control. He also voiced concern about AI systems being used for their own research and development.
The complaint says risks could grow when developers use AI models to design and improve the next generations of AI with limited human oversight. The plaintiffs argue this process, known as “recursive self-improvement,” could make advanced AI even harder to control. The filing states that the defendants know these systems do not always behave as intended and that no reliable method exists to control them or stop them from causing harm. The suit comes amid growing safety concerns following reports of autonomous AI agents allegedly taking unauthorised actions.
The groups also cited earlier warnings from AI leaders. In 2015, OpenAI CEO Sam Altman warned that AI could end humanity. In 2025, Anthropic CEO Dario Amodei estimated a 25 percent chance that AI development could lead to catastrophic outcomes. In October, Altman, Amodei and SpaceX CEO Elon Musk expressed support for a coordinated slowdown in AI development, following a warning from an Anthropic researcher who estimated a 10 percent risk of AI causing human extinction.