AI tools like ChatGPT, Gemini, Claude, and Copilot are everywhere now, and it’s natural to want to use them to make sense of a legal matter — summarizing a letter, organizing documents, or taking notes on a call. But when it comes to your legal matter specifically, a few common habits can create real problems: they can expose confidential information, and in some cases they can waive the attorney-client privilege that protects your case altogether.
This is general information to help you avoid common pitfalls — not legal advice about your specific situation. If you’re ever unsure whether something is safe to run through an AI tool, the safest move is always to ask your attorney first.
1. Communications With Your Lawyer
Don’t paste emails, letters, or messages from your attorney into a public AI chatbot to have it “summarize,” “explain,” or “translate” them for you — even if it feels like a harmless shortcut.
On February 10, 2026, a federal judge in the Southern District of New York (Judge Jed Rakoff, in United States v. Heppner) ruled that documents a defendant created by feeding information from his defense attorneys into a consumer AI chatbot were not protected by attorney-client privilege or the work-product doctrine — and that doing so may have waived privilege over the original attorney communications themselves, not just the AI-generated output. See Your AI Conversations Are Not Privileged (Jones Walker LLP). That ruling arose in a criminal case, but the court’s reasoning is not limited to criminal matters — it applies to any use of a consumer AI account to process privileged communications.
What to do instead: If something in a letter or email from your attorney is unclear, ask your attorney to explain it. That conversation stays protected; a chatbot conversation about the same letter may not.
2. Documents and Other Case Information
Don’t upload contracts, discovery materials, medical records, financial statements, or other case documents to a personal or free-tier AI account for review, organization, or summarization.
Most consumer AI tools — including free and standard individual-paid accounts — use your inputs to train their models by default, and their privacy policies generally allow them to retain that data and disclose it in response to legal process. Only enterprise-level agreements with contractual confidentiality and no-training terms are different, and your law firm’s own systems, not your personal AI account, are the ones set up with those protections in mind.
What to do instead: Send documents to your legal team the way your firm asks you to — through its secure client portal, encrypted email, or another approved channel. If you want help organizing or understanding a document, that’s exactly what your legal team is for.
3. Transcriptions and Notetaking in Videoconferences
Don’t add an AI notetaker or transcription bot (Otter.ai, Fireflies, Zoom AI Companion, or similar) to a call with your attorney without checking first — even if it’s a feature already built into your calendar or meeting account.
These tools function as an uninvited third party sitting in on what would otherwise be a private, privileged conversation, and litigation is already underway over how some of them handle recorded conversations. See Otter.ai Lawsuit Highlights Privacy & Compliance Risks of AI Note-Taking. Separately, roughly a dozen U.S. states — including California, Florida, Illinois, Maryland, and Pennsylvania — require every participant’s consent before a conversation can be recorded at all, so an AI notetaker joining a call without everyone knowing can raise its own legal problems, independent of privilege.
What to do instead: Ask your attorney before enabling any recording or transcription tool on a call, even one that seems automatic or “built in.” If notes are needed, your attorney can arrange for that in a way that protects the conversation.
4. Invoices Generated by the Law Firm
Don’t run your legal invoices through a consumer AI chatbot or a general AI-powered expense/bookkeeping tool to categorize, summarize, or “explain” the charges.
It’s easy to think of a legal bill as just an accounting record, but the line-item descriptions on a legal invoice often describe the substance of the work performed — what was researched, what strategy was discussed, what documents were reviewed. That narrative detail can itself be privileged or protected work-product information, not merely a dollar figure.
What to do instead: If you have questions about an invoice, ask your firm’s billing contact or your attorney directly. If your business uses an AI-assisted bookkeeping or expense tool, confirm with your attorney (and, ideally, your own IT or finance team) whether that specific tool has confirmed confidentiality protections before feeding it any legal billing detail.
5. Using AI to Think Through Personnel Decisions or Other Sensitive Business Matters
Don’t use a chatbot to help think through, draft, or justify a sensitive decision — like a termination, discipline, or other personnel action — as though the conversation is private. This applies whether or not a lawsuit has been filed yet.
Courts are actively sorting out whether AI chat logs are protected in litigation, and early decisions have gone both ways. In one employment discrimination case, a federal court in Michigan found that a self-represented plaintiff’s ChatGPT prompts, used to help her draft her own court filings, were protected as attorney work product. One week later, a federal court in New York reached the opposite conclusion in a very different context, holding that a criminal defendant’s AI chat logs were neither privileged nor protected (the same Heppner ruling discussed in Section 1 above). See Can Your AI Chat History Be Used Against You in a Lawsuit? (Fisher Phillips LLP). The trend across these decisions is that courts increasingly treat AI conversations as just another form of electronically stored information — similar to an email or text message — meaning they can be requested in discovery and potentially used as evidence.
That has real consequences for personnel and other sensitive business decisions. If a manager or HR representative uses an AI tool to help think through, draft, or explain the reasoning behind a termination, discipline, or similar decision, that conversation can itself become evidence later — including evidence of what the decision-maker was actually thinking or intending at the time. Employment counsel are now specifically advising employers to limit or avoid AI use for hiring, firing, and other personnel matters for exactly this reason.
What to do instead: Treat a conversation with an AI tool the same way you’d treat an email or memo you wouldn’t want read aloud in a deposition. Before using AI to help think through a termination, discipline decision, or any other action that could later be challenged, talk to your attorney first. This is a new and unsettled area of law, and how courts treat these questions may vary by jurisdiction and by the specific facts involved.
The Bottom Line
None of this means AI tools are off-limits in your life generally — it means that anything connected to your legal matter deserves a pause before it goes into one. A simple rule of thumb: if the information came from your attorney, is about your case, or describes the legal work being done for you, don’t put it into an AI tool without checking with your legal team first.
This article is general information, not legal advice, and it isn’t a substitute for guidance about your specific matter. If you have questions about what’s safe to share and how, please contact your attorney.
Business Law Southwest. Business Law That Makes Business Sense. A Slingshot Company.
