For centuries, attorney-client privilege has served as one of the most sacred protections in law. It allows clients to speak freely with their lawyers, knowing their secrets will remain confidential even in the face of government scrutiny.
But what happens when those secrets pass through machines?
The decision in United States v. Heppner illustrates, with doctrinal clarity, that attorney–client privilege does not exist as an abstract entitlement. Rather, it is contingent upon the preservation of confidentiality as a factual and legal condition precedent. Where confidentiality is compromised, whether deliberately or inadvertently, privilege ceases to operate. While the court in Heppner did not address generative artificial intelligence directly, its reasoning exposes a profound vulnerability in privilege doctrine when applied to modern technological intermediaries. In a legal ecosystem increasingly mediated by generative AI systems that process, retain, and operationalize user input beyond the immediate control of the client or attorney, the foundational assumptions underlying privilege are under strain.
United States v. Heppner
The defendant asserted attorney-client privilege over specific communications, arguing that the information was protected because it related to legal consultation. However, the government challenged this claim, arguing that the defendant had disclosed the information through channels that involved third parties, thereby destroying the confidentiality required for privilege to apply.
The central factual tension was not whether legal advice existed, but whether the confidentiality surrounding that advice had been preserved.
This distinction proved decisive.
