Do you use AI tools like ChatGPT, Claude or Deepseek to draft emails, or research everyday questions or problems? You may have even used AI tools to analyze legal documents or understand how laws apply to you or your businesses. But are those conversations private? 

AI tools are a new area courts in the US have been working to understand on how they fit into our legal framework. Courts are increasingly treating AI chat logs as discoverable electronically stored information (ESI) just like emails, Google searches or text messages. This article explains how courts can request your AI chats in civil litigation, why they are generally not confidential, and what you can do to protect yourself.

How AI Chats Become Discoverable in Litigation

Under the Federal Rule of Civil Procedure 34(a), parties may request production of documents and ESI (electronically stored information), including data  and digital records. Courts inbcluding California courts interpret ESI broadly to include emails, instant messages, platform logs, and digital records and courts have determined that AI chat histories fit within this definition. Courts have evolved to recognize internet search histories and emails as potentially discoverable, there is an emerging risk that AI inputs and responses, mimicking correspondence, may be used to show how a litigant analyzed facts and information and created documents outside of the protection of attorney-client privilege or attorney work product protections.

Discovery Is Broad

Federal Rule of Civil Procedure 26(b)(1) and California Civil Procedure permits discovery of any nonprivileged matter that is relevant to a party's claim or defense and proportional to the needs of the case. In the same was Discovery or the process of obtaining evidence and documents from a person can obtain text messages and emails, AI Chat logs can also be obtained.

AI chat logs can reveal: A party's intent, knowledge, and thought processes, Business strategies and internal deliberations and How a party analyzed facts and created documents outside of privilege protections.

(People ex rel. Dept. of Public Works v. Donovan (1962) 57 Cal. 2d 346, 355, 19 Cal. Rptr. 473, 369 P.2d 1; Wood v. Superior Court (2020) 46 Cal. App. 5th 562, 259 Cal. Rptr. 3d 798; Jasper Constr., Inc. v. Foothill Jr. College Dist. (1979) 91 Cal. App. 3d 1, 17, 153 Cal. Rptr. 767).

Why AI Chats Are Generally Not Confidential

Attorney-client privilege protects confidential communications between a client and attorney made for the purpose of obtaining legal advice. Since communications with an AI platform are not considered "privileged and confidential" because they are not between a client and an attorney there is no privileged relationship, so these chats are considered as a conversation with any other person, which can be subpoenaed by the court. 

Additionally, AI platforms like Claude explicitly disclaim providing legal advice, undermining any claim that the user sought "legal advice" from the AI.

Also Anthropic's privacy policy (for Claude) allows data collection on user inputs and outputs, use of data to train the model, and disclosure to third parties, including governmental regulatory authorities. Meaning even if you delete your chats, the server will have copies of your conversations for several years that can be subpoenaed by the court. Their terms and conditions also state that by using the platform, users consent to these terms, eliminating any reasonable expectation of confidentiality.

Recent cases have determined that when a user communicates with AI of their own volition not at your attorney's direction the AI is not functioning as the attorney's agent. So there is no confidentiality according to Federal and California law.

Work Product Doctrine: Limited Protection

The work product doctrine is meant to protect materials prepared by or on behalf of an attorney in anticipation of litigation or trial. Its purpose is to protect the mental processes of the attorney, not the client's independent work. So if you use AI and are not an attorney there is a real danger that these chats can be requested and subpoenaed by the court. 

If the you act independently, without attorney direction, work product protection does not apply and they can be requested by the court.  A client's independent internet searches or AI queries even if later shared with your attorney do not become privileged merely because they are shared.

Case Example United States v. Heppner (2026)

In this case Bradley Heppner, facing federal criminal charges, used Claude (a generative AI platform) to prepare defense strategy reports. He did so independently, without direction from his attorneys, and then shared the outputs with counsel. Even though they were shared with his attorney later, since he had made these prompts were made independent of any guidance by an attorney the FBI was able to seize 31 AI-generated documents from his home.

The court concluded that Claude is not an attorney; no privileged relationship could exist and since Anthropic's privacy policy permits data collection and disclosure to third parties, this eliminated any reasonable expectation of privacy. Sharing AI outputs with your lawyer at a later time does not make them privileged.

Practical Takeaways for Business Owners and Individuals, Assume AI Chats Are NOT Private

Since AI chats can be obtained by the court a good practice point is to treat every AI interaction as if it could become part of a lawsuit.  Avoid sharing trade secrets, client confidences, or attorney communications with public AI platforms.  If you are involved in litigation or anticipate a dispute, do not use AI independently to analyze your case. If your attorney directs you to use AI, the analysis may be different, but do not assume protection.

If you use AI in your buisness and workplace, Ensure your business's preservation policies explicitly address AI chat logs and prompt histories.

The law is evolving rapidly to catch up with technology, but the foundational principles remain unchanged: privileged communications require a confidential relationship with a licensed attorney, and work product protection exists to safeguard counsel's mental processes, not a client's independent experimentation with AI. When you input sensitive business strategies, legal theories, or confidential facts into a public AI platform, you are effectively having a conversation with a third party that retains that data, trains on it, and may be compelled by subpoena to produce it.

How the Law Office of Steve Lopez Can Help

At the Law Office of Steve Lopez, we understand the complex intersection of emerging technology and traditional legal protections. Whether you are facing a discovery request that seeks your AI communications, need to update your workplace policies to address AI usage, or require defense in a civil litigation matter, we are here to protect your rights.