Join our email list today to receive alerts, articles, invitations to events and more!
2 min read
Overview
AI is transforming legal workflows, but it can also jeopardize solicitor-client privilege when sensitive information is shared on public platforms. Communications between clients and AI tools may not be protected and could become discoverable in legal proceedings. Courts are applying traditional legal principles, reinforcing that privilege depends on confidentiality and direct legal counsel. However, privilege can still be preserved when AI is used within secure systems and under lawyer supervision. The key is balancing innovation with caution, ensuring confidentiality is never compromised.
Let’s imagine you're in the middle of a lawsuit, and you consult an AI tool about your case. For the sake of discussion, let's say it's Claude from Anthropic. Have you just waived solicitor-client privilege?
"Solicitor-Client Privilege" is a rule that protects communications between a client and their lawyer. It's intended to encourage a client to communicate candidly and in confidence with their lawyer, knowing that these communications are protected from disclosure.
Two recent US cases show how courts are treating this issue.
In United States v. Heppner (S.D.N.Y. Feb. 17, 2026), the defendant, Heppner, who was charged with a number of offences related to securities and wire fraud, consulted Claude about the charges, and the documentation of these AI exchanges was later the subject of a separate challenge by the plaintiff. The US government (the plaintiff) sought production of these records. Heppner's legal team argued they were protected by privilege. The court, in its February 17 decision, ordered the production of the documents, saying they failed to satisfy the core elements of attorney-client privilege: namely, confidentiality, communications between lawyer and client, and having the purpose of obtaining legal advice. "Attorney-Client Privilege" is the American version, and the concept is the same as "Solicitor-Client Privilege" in Canada.
As the court noted in its decision, Anthropic’s privacy policy “provides that Anthropic collects data on both users’ ‘inputs’ and Claude’s ‘outputs,’ that it uses such data to ‘train’ Claude, and that Anthropic reserves the right to disclose such data to a host of ‘third parties’…” This supported the court's conclusion that privilege did not protect the defendant's AI-generated documents.
Meanwhile, in Warner v. Gilbarco (E.D. Mich. Feb. 10, 2026), the court came to a different conclusion in a case where a self-represented plaintiff created documents using AI tools: internal drafts about her case, and legal analysis generated through ChatGPT. The defendants sought production of these documents, arguing that any privilege or protection was waived by disclosure to a "non-human third party."
The court decided in this case that the plaintiff’s AI-related materials were not discoverable, since they were protected as "work product" created for the purposes of litigation. The court reasoned that ChatGPT is not a "person" but merely a tool. Perhaps most importantly, this plaintiff was not represented by counsel, so the documents reflected her own mental impressions and strategies to assist in case preparation, putting them outside the reach of discoverability.
Each case will turn on its facts. It's important to note that the use of AI may result in the erosion of protections afforded by solicitor-client privilege. AI platforms must be used cautiously and with the advice of counsel.
Contact Richard Stobbe or any member of our Artificial Intelligence Group for guidance on the legal risks of using AI tools, including impacts on solicitor-client privilege, confidentiality, and data handling practices.
