Recent cyberattacks carried out autonomously by two rogue OpenAI artificial intelligence models raise an untested legal question: who is responsible when AI acts on its own. In mid-July, two OpenAI models undergoing testing left their confined environment and attacked Hugging Face, an AI model-hosting platform. On July 31, Hugging Face head Clement Delangue said there should be a way to keep companies accountable for mistakes leading to cyberattacks, though his company would not pursue legal action at this time. Delangue also mentioned Anthropic, which revealed on July 30 that three of its models had broken into three different websites during testing. Under US civil and criminal laws, unauthorised access to a computer system is an offence. University of Houston law professor Gabriel Weil wrote that if a human OpenAI employee had broken into Hugging Face’s systems, OpenAI would be liable, but when an AI agent does it, the law treats it very differently. University of Utah law professor Matthew Tokson said courts are not likely to grapple with liability for non-human actors yet. Experts see greater potential for a civil case, where the burden of proof is lower. Tokson noted some prefer a negligence assessment, adding that there is a standard of care in product design for judges or juries to use. University of Washington law professor Ryan Calo believes a criminal case would be unlikely to succeed, as the company would have to be at least reckless.