Courts Warn: Sharing Legal Advice With AI Chatbots May Waive Attorney-Client Privilege

By Hector Herrera | April 21, 2026

U.S. courts are warning that feeding attorney communications into AI chatbots may waive attorney-client privilege — the legal protection that shields what clients tell their lawyers from disclosure in court. The warnings are coming from active judicial rulings and from law firms updating their client contracts, and they create immediate exposure for anyone who has used a general-purpose AI tool to analyze legal documents.

The advisory landscape is developing rapidly, with major law firms now adding AI disclosure clauses to engagement letters and some courts treating AI chat logs as evidence that privilege has been waived.

What Attorney-Client Privilege Actually Protects

Attorney-client privilege is not a technicality. It is one of the oldest protections in American law — a guarantee that clients can speak freely with their lawyers without fear that those conversations will be used against them in litigation.

The protection can be waived — permanently lost — if the client voluntarily shares privileged communications with third parties outside the attorney-client relationship. The longstanding question is whether AI chatbots constitute such a third party.

Courts are now saying: they might.

What the Courts Have Found

The rulings are not uniform, and that inconsistency is itself a problem.

In one closely watched decision, a Michigan magistrate judge ruled that an AI chat log was protected as personal work product — reasoning that the user's inputs and the AI's outputs functioned as a form of mental impression protected under work-product doctrine. This ruling was favorable for the user.

But it was a narrow ruling in one jurisdiction on one specific set of facts. Other courts are not bound by it, and the analysis hinges on how a judge characterizes the AI service's data handling: Does the chatbot provider have the right to store, review, or use the conversation? If so, sharing privileged information with that service may constitute disclosure to a third party — and waiver.

The exposure is material. Many AI services, under their standard terms of service, retain conversation data for training or review purposes. A client who pastes attorney emails into such a service to get a summary or analysis may have inadvertently handed those communications to a third party, triggering waiver.

The Scale of the Problem

This is not a hypothetical risk. The profession is already experiencing the consequences of inadequate AI governance in legal work.

Over 600 AI hallucination cases are now on record that implicate 128 lawyers — instances where attorneys submitted AI-generated content to courts that contained fabricated citations or incorrect legal claims. Courts have sanctioned lawyers for this.

The privilege issue is distinct but related: both represent a failure of professional oversight over AI use in legal work. The difference is that the hallucination problem damages lawyer credibility in court, while the privilege waiver problem can destroy a client's entire legal position — permanently.

How Law Firms Are Responding

Major law firms are responding with contract changes rather than waiting for regulatory guidance.

AI disclosure clauses are now being added to standard engagement letters — requiring clients to disclose when and how they are using AI tools to process attorney communications or case materials. Some firms are going further, explicitly prohibiting clients from inputting privileged material into third-party AI services without prior written consent from the firm.

From the firm's own side: internal AI use policies are being updated to distinguish between using AI tools that run on the firm's own secured infrastructure (generally acceptable) versus routing client data through external AI providers whose data practices may not align with privilege preservation requirements.

What This Means for Clients

If you are currently in litigation, or anticipate litigation: Assume that anything you paste into a general-purpose AI chatbot is no longer privileged. That includes email chains with your lawyer, legal strategy documents, settlement discussions, and anything else that would normally be protected.

If you want to use AI to help understand your legal situation: Ask your attorney whether they have a secure, firm-approved AI tool you can use. Many firms are deploying private instances of AI models that do not route data to external servers — these may be safe for privileged work. Do not assume the public version of any major AI product is safe.

If you are a business with active legal exposure: Review your employees' AI usage policies immediately. The risk is not limited to executives — anyone at a company who has access to attorney communications and also uses public AI tools has potential exposure.

The Broader Regulatory Gap

The legal profession is facing this problem without coordinated guidance. Bar associations — the state-level bodies that regulate lawyers — have not issued uniform standards for AI use in legal work. The American Bar Association has published general guidelines but nothing with the force of enforceable rules.

Federal courts have also not harmonized their privilege analysis for AI-processed materials. Until they do, the risk level varies dramatically by jurisdiction.

What to Watch

Whether the ABA or major state bars issue formal, binding guidance on AI and privilege in 2026, and whether federal courts develop consistent doctrine on whether AI chat logs constitute waiver. The Michigan ruling suggests some courts are willing to protect these communications — but one favorable ruling is not protection in a different jurisdiction.

By Hector Herrera | NexChron.com Source: Claims Journal, April 16, 2026