Artificial intelligence tools can be helpful for everyday tasks, but they can create serious problems when you are involved in a legal matter. Many people now use chatbots to summarize documents, draft messages, organize timelines, or ask what they should do next. The problem is that information entered into some AI services may not be private in the way people assume. In some situations, those inputs may later be discoverable, may not be protected by the attorney-client privilege, and may be used by the other side or reviewed by a court.
The safest rule is simple: do not put case facts, strategy, private documents, medical records, settlement discussions, or communications with your lawyer into a public or consumer AI service unless your attorney specifically tells you it is safe to do so.
Why this matters
Many AI products are built to collect, store, and process whatever a user types or uploads. Depending on the platform and its terms of use, the provider may retain that information, review it, use it to improve the system, or disclose it in response to legal process. That creates risk for anyone involved in a lawsuit, insurance claim, employment dispute, criminal investigation, or business dispute.
California’s State Bar has warned lawyers that they must not put client confidential information into generative AI systems that lack adequate confidentiality and security protections, and that lawyers should review the provider’s terms of use and data practices before using such tools. The same caution should matter to clients. If even lawyers have to be careful, clients should be extremely cautious before putting sensitive facts into a public AI tool.
What can go wrong
1. Your AI chats may not be privileged
People often assume that typing something into a chatbot is similar to telling it to a lawyer or storing it in a private note. That is not necessarily true. In 2026, a federal judge in New York ruled that a criminal defendant’s communications with a consumer AI assistant were not protected by the attorney-client privilege or work-product doctrine. The court treated the chatbot as a third party, not as counsel, and required production of the materials. The lesson is not that every AI interaction will always be discoverable. The lesson is that courts may treat public AI platforms as third parties, especially when the platform’s own terms do not promise strict confidentiality.
2. Your inputs may be discoverable
Even outside the privilege issue, what you type into an AI service may become relevant evidence. If you use AI to draft a statement, reconstruct events, test legal arguments, summarize your medical history, or ask how much your case is worth, those prompts and outputs may later be requested in discovery depending on the claims, defenses, and court rulings in your case. That is especially true if the AI tool was used to create a document that is later produced, relied on, or discussed.
3. You may accidentally reveal harmful facts or inconsistencies
Clients sometimes speak more freely to a chatbot than they would in an email. They may guess at dates, describe events loosely, or ask hypothetical questions that do not accurately reflect what happened. Those statements can create problems if they are later preserved and compared to sworn testimony, medical records, social media posts, or discovery responses. Even an innocent attempt to “organize your thoughts” with AI can produce wording that is inaccurate, incomplete, or misleading.
4. AI can generate wrong information
AI systems are known to generate false or invented information, sometimes in a very confident tone. That problem is called hallucination. If a client relies on AI instead of their attorney, they may take the wrong steps, miss deadlines, contact the wrong people, or misunderstand their rights. AI should never replace legal advice from a lawyer handling your matter.
What clients should not put into AI tools
- Messages to or from your attorney
- Facts about your accident, injury, claim, or lawsuit that are not already public
- Medical records, bills, diagnoses, photographs, or treatment summaries
- Settlement demands, settlement offers, or your bottom-line number
- Documents produced in discovery or marked confidential
- Employment records, disciplinary history, personnel files, or wage records
- Names of witnesses, their statements, or what you think they will say
- Your guesses about who is at fault, what really happened, or how to “beat” the case
- Draft declarations, discovery responses, timeline summaries, or testimony outlines
Safer alternatives
If you need help organizing information for your lawyer, use ordinary methods your lawyer has approved. That may mean writing notes on paper, using a secure client portal, sending information directly to your lawyer, or using a firm-approved intake form or document-sharing system. Ask before using any technology you are unsure about.
If you want to summarize a long set of records or create a timeline, do not upload it to a public AI tool first. Send it to your lawyer and ask how they want it handled. In some matters, a law firm may use approved internal tools with better security protections. In other matters, the right answer may be not to use AI at all.
A practical rule for clients
Before typing anything case-related into a chatbot, ask yourself one question: would I be comfortable if a judge, opposing counsel, an insurance company, or the other side read this later? If the answer is no, do not put it into the tool.
What we tell clients
At Impact Attorneys, we recommend that clients do not use public AI services to discuss their case, upload documents, evaluate settlement, summarize medical treatment, or test legal theories without first speaking with us. A chatbot is not your lawyer. It may not be confidential, and what you type may create avoidable problems.
The better approach is to communicate directly with your attorney. If you are unsure whether a tool is safe, ask first.
Frequently asked questions
Can the other side really get my AI chats?
Possibly. It depends on the platform, what you entered, how the information was used, and how the court rules. But there is enough risk that clients should act cautiously and assume that anything important entered into a public AI tool could become an issue later.
Does attorney-client privilege automatically protect AI chats about my case?
No. Privilege protects confidential communications with your lawyer for purposes of legal advice. A consumer AI platform is not your lawyer, and courts may treat it as a third party.
What if I already used AI to help with my case?
Tell your lawyer promptly and honestly. Do not delete anything unless your lawyer tells you to do so. Your lawyer needs to know what was entered, what documents were uploaded, what the system produced, and whether you relied on it.
Can I use AI for general, non-case-related questions?
General, non-sensitive questions are different from entering case facts or private documents. But once the question becomes tied to your legal matter, your safest course is to stop and ask your lawyer.
Bottom line
AI tools may seem private, convenient, and harmless. In a legal matter, they are not always any of those things. Inputs may be stored, shared, or later requested. Some courts may refuse to treat those communications as privileged. And AI can generate incorrect or damaging language that creates unnecessary risk.
If you are involved in a claim or lawsuit, the safest practice is to keep case-related information off public AI platforms and communicate directly with your attorney instead.
Disclaimer. This article is for general informational purposes only and is not legal advice. Every case is different. If you have questions about your own matter, speak with a lawyer about the specific facts of your case.