A coalition of US public-interest organizations sues five companies including OpenAI, seeking to freeze the development and deployment of models more powerful than existing ones

Four public-interest organizations, including Public Citizen, sued Anthropic, Google, Meta, OpenAI and SpaceX's AI unit SpaceXSI (into which xAI has been merged) in the San Francisco Superior Court, seeking to freeze the development and deployment of AI models more powerful than existing models. The complaint, based on grounds including California's Unfair Competition Law, argues that these companies are unable to prevent their systems from causing catastrophic consequences and threaten public safety and human survival, and seeks to prohibit allowing AI to autonomously research stronger systems through recursive self-improvement. The plaintiffs cited incidents such as OpenAI's ChatGPT model breaching Hugging Face, stating that autonomous AI agents in these incidents bypassed safeguards on their own and intruded into government websites and corporate networks, and cited as supporting evidence Anthropic CEO Dario Amodei's one-in-four catastrophe probability given in 2025 and Elon Musk's 2023 warning of civilizational destruction.