Why now
Courts are drawing the line on client AI use.
Two federal courts, the same month, reached opposite results on whether client AI use is discoverable. The difference between them is the difference CounselRoom is built on.
Background
Clients do not wait for permission.
They paste demand letters, contracts, and their own account of a dispute into consumer chatbots, because the tools are helpful and free and always available. Until recently, nobody knew how courts would treat those conversations when the other side asked for them in discovery. Now we are finding out.
United States v. Heppner
A represented defendant used the consumer version of a public AI tool, on his own initiative, to generate roughly thirty documents mapping out his defense. He then shared them with his lawyers, who had not directed the work. When the other side asked for those documents, the court ordered them produced. The service's consumer terms let the provider retain what he typed, use it for training, and disclose it to third parties, including the government.
The court held that neither attorney-client privilege nor the work product doctrine applied. The tool is not a lawyer. Its consumer terms made the channel anything but confidential. And the work was done on the client's own initiative, not at counsel's direction, so it reflected no lawyer's judgment. The court added that an enterprise deployment with real confidentiality terms would present a materially different question.
What the court examined
- Whether an attorney directed or supervised the AI use
- Whether the channel was confidential, including the provider's own data terms
- Whether the material reflected a lawyer's judgment, or was prepared at a lawyer's direction
Warner v. Gilbarco, Inc.
One week earlier, a federal court in Michigan went the other way, and the contrast is the whole point. A self-represented plaintiff had used a consumer AI tool to help prepare her own filings. The defense moved to compel everything about that use. The court refused. Running material through an AI tool was not disclosure to an adversary, so it did not waive work product, and a person representing herself is both the party and her own advocate.
“Generative AI programs are tools, not persons.”
Two courts, the same month, opposite results. The difference was not whether AI was used. It was who was holding the tool. Warner was her own advocate, so her preparation stayed protected. Heppner was represented, but worked alone and outside his representation, so his did not. The line the courts are drawing runs straight through the middle of that distinction, and it is the line CounselRoom is built on.
What this means for intake
The question is not whether your clients use AI.
It is whether their AI use happens in a public tool their opponent can reach in discovery, or inside the representation, directed and reviewed by their lawyer.
CounselRoom is built around the factors the courts examined in Heppner and Warner: firm-controlled access, attorney direction and supervision, and a record that shows both. It is designed to support privilege and work-product claims. No software can guarantee them, and anyone who says otherwise is selling something.
This page summarizes public court decisions in plain language, for general information. It is not legal advice. How these decisions apply to any matter depends on the jurisdiction and the facts, and the law in this area is developing quickly.
Bring intake inside the representation.
CounselRoom is in an invite-only pilot for small and midsize firms.