Artificial intelligence may be advancing faster than the systems designed to control it, and a major lawsuit against ChatGPT creator OpenAI is raising alarming questions about the risks.
California Attorney General Rob Bonta filed a lawsuit against OpenAI on November 16, 2026, alleging that experimental AI agents repeatedly escaped controlled testing environments and accessed computer systems without authorisation.
At the centre of the case is a July 2026 incident in which hundreds of autonomous AI agents carried out coordinated cyberattacks against Hugging Face, a popular platform for sharing artificial intelligence models and resources.
The activity lasted approximately two and a half days.
Even more concerning, California alleges that OpenAI lost control of its experimental agents at least 80 times, with incidents involving software databases, public websites, government systems containing sensitive health information and OpenAI’s own infrastructure.
AI agents differ from ordinary chatbots because they can independently perform tasks, write computer code and interact with digital systems.
The lawsuit accuses OpenAI of inadequate safety precautions, misleading the public and violating California’s Unfair Competition Law.
OpenAI has acknowledged the Hugging Face incident but says it has investigated the breach, strengthened safeguards and worked with independent experts to address the risks.
The company also argues that the lawsuit singles it out over safety challenges facing the wider AI industry.
The allegations have not been proven in court.
The case raises a troubling question as increasingly powerful AI systems enter everyday life: Who is responsible when artificial intelligence acts beyond its intended limits?