The Question Businesses Should Be Asking
Hey Chat, how do I dump my business partner? Claude, how do I break this contract? Recent court cases have demonstrated that these AI prompts and chatbot responses, fired off from a laptop in the privacy of someone’s home, may not be private simply because they involve legal-sounding questions. This raises questions many businesses have not fully considered: what happens when employees ask public AI tools about disputes, liability, contracts, investigations, or regulatory issues?
From Courtroom Theory to Business Reality
As we noted in our recent Daily Journal article, “The Ghost in the Courtroom: From AI Chatbots to Agentic Litigation,” attorney-client privilege depends on confidential communications with a lawyer. Sorry, Claude. We explored how AI may reshape litigation, privilege, discovery, and the evidence trail. This follow-up focuses on one immediate risk for businesses: what happens when employees use public AI tools to discuss disputes, contracts, investigations, or regulatory problems before counsel is involved?
AI Prompts May Not Be Private
Chatbots may cheer you on, but they can’t always keep your secrets. If employees use AI, their prompts, uploads, summaries, and outputs could all be discoverable. Even if an employee is asking a legal question, attorney-client privilege is not invoked. Privilege is connected to a licensed lawyer.
Why Counsel’s Role Matters
The lawyer may use AI tools such as ChatGPT or CoCounsel while providing legal advice and working in anticipation of litigation. A lawyer can use AI to generate options, interview or organize information, draft documents, and compare lower and higher risks. All of that can be protected within the attorney-client relationship. That would not always hold true for a private individual’s unrepresented use of the same technology.
Regulatory Risk Raises the Stakes
Public-use AI bots offer the same exposure in regulatory matters involving FINRA, the SEC, or the FTC. If someone communicates with an AI system without counsel and later receives a regulatory demand for documents, those AI records generally don’t become privileged simply because the questions asked were legal.
What Could Become Evidence
Bottom line? Searches, prompts, chat logs, and follow-up exchanges entered by an individual into ChatGPT or similar tools may be discoverable and admissible in civil litigation. Plaintiffs’ attorneys could seek those records as part of document requests, particularly where the user asked questions about disputes, liability, or avoiding payment.
Practical Steps Before a Dispute Starts
- Don’t input sensitive dispute facts into public AI tools without legal guidance
- Don’t upload contracts, emails, board materials, investigation notes, or regulatory correspondence into public AI tools without approval from your lawyer
- Involve counsel early when AI will be used for litigation, investigation, or regulatory strategy
- Create an internal AI-use policy that addresses privilege, confidentiality, record retention, and discovery
- Train employees on what NOT to put into AI tools
- Preserve AI records when litigation or regulatory inquiries are reasonably anticipated
From Chatbot Logs to Agentic Litigation
Today, the concern is prompts and chatbot logs. Tomorrow, we’ll likely be facing autonomous systems making decisions, entering transactions, managing obligations, and creating discoverable evidence trails. The floor is shifting beneath our feet at lightning speed and human lawyers still matter for businesses and legal teams navigating this ever-changing landscape.



