More than ever before, clients are using artificial intelligence (AI) prior to approaching their attorneys for legal advice. Doing so proves to have many advantages. First, it allows clients to get a basic understanding of legal issues relevant to their situation. Second, armed with this information, clients and attorneys are likely to be engaged in productive conversations from the beginning of the attorney-client relationship. Third, clients are able to proactively ask their attorneys relevant questions or flag areas that are of particular concern. Most importantly, perhaps, AI use could lead to cost savings for clients, as AI use can give clients a head start prior to meeting with their attorneys.

However, clients also should be aware of potential risks and concerns with this type of use of AI. Aside from receiving incorrect or outdated information, using AI prior to approaching their attorneys risks waiving attorney-client privilege (ACP). Every attorney-client relationship is premised on ACP, a rule of evidence that protects confidential communications between an attorney and their client that relates to the client’s seeking or receiving legal advice or services. If ACP applies, the communication is protected from disclosure in discovery in the event of litigation. This means opposing counsel cannot be privy to what a client candidly discussed in confidence with their attorney, including strategies or unfavorable facts that may benefit the other side. ACP allows attorneys to effectively and diligently represent their clients while protecting clients’ confidential information.

ACP, however, is not absolute, and clients can, whether expressly or inadvertently, waive it. A more fundamental risk also exists: ACP may never attach to a client’s communications with free AI platforms, such as ChatGPT, Google Gemini, or Claude, because some or all of the elements of ACP fail. For example, prior to retaining counsel, if a user uploads a contract, unpaid invoices, email exchanges, and a settlement offer to ChatGPT, asks it to analyze those materials, and generates a strategy memorandum, will ACP attach and protect the chat history or the memorandum from discovery? The answer, unfortunately, is no.

• The communications are not “confidential.” Users have no reasonable expectation of privacy in their interactions with free versions of AI chatbots. In their terms of service, such platforms often disclaim confidentiality, stating that they use user inputs and system outputs to train their models, even in instances where a user chooses to opt out. For example, Claude’s terms of service state: “We may use [user interactions and generated responses] to provide, maintain, and improve the Services and to develop other products and services, including training our models, unless you opt out of training through your account settings. Even if you opt out, we will use Materials for model training when: (1) you provide Feedback to us regarding any Materials, or (2) your Materials are flagged for safety review to improve our ability to detect harmful content, enforce our policies, or advance our safety research.”

• AI tools are not “communications between an attorney and their client.” AI tools are not attorneys. In fact, there is no attorney involved at all in a scenario where a user is using AI platforms prior to engaging with an attorney. The discussion of any legal issues between two non-attorneys, such as an unrepresented user and a free AI chatbot, are not protected by ACP, as there is no attorney-client relationship present.

• A user isn’t “seeking or receiving legal advice” when using an AI platform of their own volition. A user acts on their own volition when they use an AI tool prior to engaging with an attorney. Therefore, when Counsel is not suggesting or directing a person’s  use of an AI tool, a court is likely to rule that user isn’t “seeking or receiving legal advice.” (The court in United States v. Heppner, No. 25-cr-00503-JSR (S.D.N.Y. Feb. 17, 2026), noted that this analysis might differ if counsel had directed the AI use because the platform could then arguably function as an agent of counsel.) AI tools often disclaim providing legal advice at all. Moreover, providing the chats or generated responses or materials to counsel after the fact does not suddenly make those documents privileged.

Practical Takeaways

Given these risks, how can users best use AI prior to seeking legal advice while keeping in mind the risk of waiving privilege?

• Do not use free AI platforms that do not have an obligation to keep your information confidential. Use AI tools that have explicit confidentiality agreements, such as enterprise AI tools, rather than public tools, such as standard ChatGPT.
• Review the terms of service for any AI platform before use. Many public platforms explicitly state that user inputs may be used to train models or may be accessed by platform employees, which could constitute third-party disclosure.
• Ensure that AI usage involving legal questions is separate from AI used for general business operations.
• After retaining an attorney, use AI to prepare for meetings with attorneys. This may include asking AI to prepare a list of documents to be gathered, a template for a timeline of events to fill out, or general questions to start the conversation regarding an issue.
• If clients have already used public AI platforms to discuss legal matters, they should disclose that use to their attorneys. Their attorneys can assess the potential impact on privilege and develop strategies to address any exposure.
• After retaining and discussing with an attorney, only use AI tools that have explicit confidentiality agreements and as directed by counsel.

Conclusion

AI can be a valuable first step for clients seeking to understand their legal options, but it is not a substitute for confidential legal advice from counsel — and careless use can compromise the privileged relationship that clients rely on. By treating public AI as a general research tool rather than a confidential advisor and by involving their attorneys early (preferably before entering any confidential or fact-specific information), clients can capture the benefits of AI while preserving privilege and confidentiality.

This article is provided for general informational purposes only and does not constitute legal advice. Readers should consult counsel regarding the application of ACP to their specific circumstances.