Writing in NLJ this week, Mary Young of Kingsley Napley examines how traditional principles of misrepresentation struggle when applied to large language models such as ChatGPT. Because AI systems are not legal persons, claimants must instead seek to attribute liability to developers, operators or companies deploying the technology.
Young points to the Canadian Air Canada chatbot case, where the airline was held responsible after false information appeared on its website, despite arguments that the chatbot operated autonomously.
The article also explores whether software developers could owe duties analogous to trustees or company directors, drawing on Tulip Trading v Van Der Laan. Yet proving fraudulent intent remains difficult where AI systems generate inaccurate information without human involvement, making successful claims an uphill struggle.
