Client Use of AI Transcription Tools Can Break Attorney-Client Privilege

Client Use of AI Transcription Tools Can Break Attorney-Client Privilege

Lawyers face a new challenge as clients increasingly turn to AI-powered transcription and note-taking tools to record conversations with their attorneys. While convenient, this practice can inadvertently destroy the attorney-client privilege that protects confidential communications, according to recent court rulings and ethics guidance.

The issue centers on how generative AI systems operate. These platforms typically record user inputs to train their models, often making it a prerequisite for using the service. When clients feed details of their legal matters into these systems, they may be sharing privileged information with third parties—the AI companies themselves.

A landmark federal ruling

In United States v. Heppner, the Southern District of New York permitted disclosure of a defendant's communications with a publicly available AI platform about his pending criminal case. The court found that all three elements required for attorney-client privilege were absent: the communications weren't between client and attorney, they weren't kept confidential given the AI's data collection practices, and they weren't made for the purpose of seeking legal advice since the program wasn't a lawyer.

The ruling also found that work-product protections didn't apply because the communications weren't prepared by or at the direction of counsel. This decision establishes that under normal circumstances, client communications with AI about their cases will be subject to disclosure.

Why it matters

This creates a practical dilemma for legal practice. Clients who unknowingly waive privilege by using AI transcription tools force their attorneys into difficult positions—potentially having to disclose factual analysis, legal advice, and settlement positions to adversaries. The resulting discovery disputes waste time and money while undermining case strategy.

The problem extends beyond privilege. Collecting AI-stored communications for disclosure isn't straightforward or inexpensive, given how these systems store data. Lawyers may face motion practice to limit disclosure, running up fees and delaying case resolution.

Ethical obligations for attorneys

The New York Rules of Professional Conduct require lawyers to make reasonable efforts to prevent inadvertent disclosure of confidential information. This duty extends to preventing clients from accidentally breaking privilege.

The New York City Bar Association recently issued formal guidance recommending that attorneys advise clients of the disadvantages of recording calls with AI tools. While not strictly mandatory, ethics experts consider it best practice to address these risks at the outset of representation, ideally in engagement agreements.

Lawyers have options: they can prohibit clients from recording conversations entirely, or they can work with clients to identify AI configurations that protect confidentiality, such as private instances with custom security settings. Either approach requires clear communication and documentation in engagement letters.

The fundamental principle remains unchanged even as technology evolves: lawyers must ensure clients understand the risks of their actions and make informed decisions about their representation. As AI tools become ubiquitous, that duty now includes explaining how these conveniences can compromise legal protections.

These details were first reported by the New York State Bar Association in an Attorney Professionalism Forum column by Vincent J. Syracuse, Alyssa C. Goldrich, Jean-Claude Mazzola, and Adam Wiener.