Information shared with AI tools like ChatGPT or Claude is not protected by attorney-client privilege and may be discoverable in court [1].
This lack of legal protection creates a significant risk for businesses and individuals who use generative AI to draft legal strategies or analyze sensitive disputes. Because these interactions are not privileged, opposing counsel could potentially access these logs during the discovery phase of a lawsuit.
Kristina Subbotina, a startup lawyer and founder of Lexsy, said that AI systems are not lawyers [1]. Consequently, they cannot claim the legal privilege that traditionally protects communications between a client and a licensed attorney [1].
Attorney-client privilege is designed to encourage full and honest communication between a legal professional and their client. This protection ensures that strategic discussions, and admissions of fact, remain confidential. However, since an AI chatbot is a software product and not a member of the bar, the legal threshold for privilege is not met [1].
Users often treat AI interfaces as private consultants or sounding boards for legal dilemmas. Subbotina said that this behavior could lead to the creation of a digital paper trail that serves as evidence against the user in litigation [1].
Companies integrating AI into their workflows may need to establish strict guidelines regarding the types of data entered into these systems. Without such guardrails, sensitive internal deliberations regarding legal risks could become public record if a court orders the production of AI prompt histories [1].
“AI systems are not lawyers”
The intersection of generative AI and legal discovery highlights a gap in current law where users mistake technical privacy for legal privilege. While a company's terms of service may promise data privacy, those terms do not override a court's power to subpoena records. This creates a new category of evidentiary risk for any professional using AI to process sensitive or potentially litigious information.
