Think Before You Chat: The Hidden Legal Risks of Using AI Before Calling a Lawyer
When a legal problem lands in your lap, the instinct to open a chatbot and start typing is completely understandable. AI legal advice is fast, free, and available at midnight when you cannot sleep. But what feels like a smart first step may actually be creating evidence against you, waiving your legal protections, and putting your case at serious risk before it even begins.
Kelleher + Holland is a well versed law firm with experience dealing with AI-related legal issues. Here is what you need to know before you type a single word into any AI platform.
Your AI Conversation May Not Be Private
This is the most important thing to understand about using AI for legal questions: most public AI platforms are not confidential. When you describe your situation to ChatGPT, Claude, Gemini, or any consumer-grade AI tool, that conversation may be stored, used to train future models, shared with third-party vendors, or, in some circumstances, disclosed to government authorities.
When you input sensitive facts, strategies, or personal details into a public AI chatbot, you are potentially sharing that information with the world. And courts are starting to take notice.
In 2026, a federal judge in the Southern District of New York ordered a defendant to turn over 31 documents he had generated using Anthropic’s Claude while preparing his legal defense. Judge Jed Rakoff ruled in United States v. Heppner that those AI-generated materials were not protected by attorney-client privilege or work product doctrine. The reasoning was straightforward: the defendant had used a consumer AI tool, not a confidential channel with his attorney, and the platform’s own terms of service acknowledged it could not guarantee confidentiality.
That same year, the Delaware Chancery Court found in Fortis Advisors v. Krafton that a CEO’s ChatGPT prompts and AI-generated strategy outputs were discoverable evidence in corporate litigation. The court did not treat AI-assisted communications any differently from an email or a memo. What goes into an AI platform can come out in discovery.
Attorney-Client Privilege Does Not Cover Your AI Chats
The attorney-client privilege is one of the most powerful protections in the legal system. It shields confidential communications between a client and their attorney from disclosure in legal proceedings. But three things must exist for privilege to apply: a lawyer, a client, and a confidential communication.
AI is not a lawyer. There is no licensed attorney on the other end of your chat window. There is no fiduciary relationship, no professional duty of confidentiality, and no privilege. When you describe your legal situation to an AI chatbot, you are not in a privileged conversation. You are typing into a third-party platform, and that information does not receive the protection that a conversation with your actual attorney would. This is not a technicality. It is a fundamental gap in protection that can directly harm your case.
Work Product Protection Usually Fails Too
You may have heard of the work product doctrine, a related legal protection that shields materials prepared in anticipation of litigation. In theory, notes you prepare about your own case could qualify. In practice, AI-generated work product from consumer tools is increasingly failing this test in court.
Courts applying the work product doctrine look at whether materials were prepared under the direction of counsel, with a reasonable anticipation of litigation. When you independently type facts and strategies into a public AI tool without any attorney involvement, those materials rarely meet the standard. As Greenberg Traurig warned in a 2026 client alert, AI-generated documents can undercut confidential communications and expose clients to discovery in ways they never anticipated.
What You Should Do Instead
The answer is simple, even if the temptation to do otherwise is real. Contact an attorney before you start researching, drafting, or brainstorming your legal situation using AI. Once you establish a relationship with counsel, your attorney can advise you on whether any AI tools are appropriate for your matter, which enterprise-grade or privacy-focused platforms may be permissible, and how to communicate about your case without creating discoverable material.
A few practical steps worth taking right now:
- Do not input facts, names, dates, strategies, or sensitive details into any public AI platform before speaking with an attorney.
- Do not assume any AI chat is private. Read the platform’s data and privacy terms. Most consumer AI tools explicitly disclaim confidentiality.
- Do not treat AI legal research as a substitute for legal advice. AI can and does produce inaccurate information, cite non-existent cases, and misstate the law in ways that are not always obvious.
- Do assume that anything you type into a public AI tool could, under the right circumstances, be produced in litigation.
AI Is a Tool. It Is Not Your Lawyer.
There is nothing wrong with using AI tools in your daily life. But when you have a potential legal problem, including a business dispute, a personal injury, a family law matter, a criminal investigation, or a contract issue, the stakes are different. What you say, where you say it, and to whom you say it all matter.
The attorneys at Kelleher + Holland understand how to navigate legal matters in an era when the line between helpful technology and harmful disclosure is not always obvious. We can help you understand what is safe to use, what to avoid, and how to protect your rights from the very first conversation.
Think before you chat. Then call us. Schedule a Consultation with our team today regardless of your legal matter.
Frequently Asked Questions
Q: Can my AI conversations be used against me in court?
A: Yes. Courts have already ruled that AI-generated documents and chat outputs can be discoverable in litigation. In United States v. Heppner (SDNY, 2026), a federal judge ordered disclosure of Claude-generated materials, finding they were not protected by attorney-client privilege or the work product doctrine.
Q: Does attorney-client privilege apply to conversations with AI chatbots?
A: No. Attorney-client privilege requires a licensed attorney, a client, and a confidential communication. AI chatbots are not attorneys and have no duty of confidentiality. Any information you share with a public AI tool is outside the scope of attorney-client privilege.
Q: Are AI chatbots like ChatGPT or Claude confidential?
A: Generally, no. Most consumer AI platforms state in their terms of service that they may store your inputs, use them to improve their models, or share data with third parties. Inputting sensitive legal information into these platforms creates a serious confidentiality risk.
Q: What is the work product doctrine and does it protect AI-generated notes?
A: The work product doctrine protects materials prepared in anticipation of litigation under the direction of counsel. AI-generated materials created independently, without attorney oversight, often fail to qualify for this protection. Courts apply traditional evidentiary standards to AI-assisted documents, just as they would to emails or handwritten notes.
Q: What should I do if I already used AI to research my legal issue?
A: Contact an attorney promptly. An attorney can assess what, if anything, you shared and advise you on the implications. Do not continue inputting additional facts or strategies into AI platforms until you have spoken with counsel.
Q: Is there a safe way to use AI when dealing with a legal matter?
A: Some enterprise-grade AI tools with strong contractual privacy protections and no model-training clauses may be appropriate in certain circumstances. However, any use of AI in connection with a legal matter should be discussed with your attorney first. General-purpose consumer AI platforms are not appropriate for sensitive legal matters.
—
This post is for informational purposes only and does not constitute legal advice. Contact our team for guidance specific to your situation.