The Hidden Risks of AI

Former NFL player Darron Lee desperately needed legal advice. His girlfriend had suffered severe injuries that he later characterized as the result of an accident. As she lay unresponsive, he consulted ChatGPT, hoping for help with a cover story for the injuries he had inflicted. 

He asked the artificial intelligence tool how to “handle someone nonresponsive” without “police trouble.”(1) ChatGPT provided advice on how to cover up the crime scene and avoid prosecution. During a hearing that followed, his conversation history with ChatGPT was admitted as evidence against him. Unfortunately for Mr. Lee, his case serves as a reminder of the privacy issue facing those who rely on AI for legal guidance.

Since the release of ChatGPT in 2022, AI has proliferated throughout society and the legal industry. Claude, Gemini, CoPilot, and others are widely used by the public for daily tasks, including answering legal questions. Lexis’ Protégé and Westlaw’s CoCounsel are popular tools for use by attorneys. Consumer use of AI has grown exponentially. Utilization by attorneys has likewise surged, reportedly doubling year-over-year since 2022.(2)  An increasing number of attorneys and clients now report using AI in connection with their legal matters(3).

While Darron Lee’s example is an extreme and dramatic one, it is illustrative of one of the major pitfalls of relying on AI for legal advice: a lack of privacy.

Client Use of AI

It may be tempting for people seeking legal services to start out by asking their favored AI chatbot some initial questions. While this can be helpful, AI use in certain situations can severely harm a client’s case. A recent decision held that a client’s conversations with the AI service Claude regarding his defense strategy could be admitted as evidence against him, because attorney-client privilege requires “a trusting human relationship. . . with a licensed professional.”(4) Because AI are not licensed attorneys, conversations with such AI services are vulnerable to disclosure.

Even pre-litigation use of AI remains controversial. A Delaware court considered AI chats used as advice prior to the onset of litigation as valid for discovery during the resulting lawsuit.(5) The defendant used ChatGPT to formulate a business plan which later resulted in litigation. This ruling pushes the potential discoverability of AI communications earlier in the dispute timeline, before either party anticipated litigation. 

While AI may allow clients to conduct research on their legal issues, the lack of privilege afforded to such communications means that people choosing to utilize AI tools for legal questions should do so carefully, with the understanding that their communications with their AI systems may not be protected by attorney-client privilege and may ultimately be subject to discovery by opposing parties, even if such AI use predates litigation.

Attorneys and AI

Legal professionals are increasingly utilizing AI for specific tasks within their practices. These include legal research, document drafting, document review, and predicting outcomes. The use of AI within law firms has resulted in increased efficiency, with Thomson Reuters estimating that use of AI could free up approximately 240 hours per year per legal professional.(6) However, the waters of AI use by attorneys remain murky.

A major concern facing attorneys who utilize AI is client privacy. The Rules of Professional Conduct require attorneys to “act competently to safeguard information relating to the representation of a client against unauthorized access by third parties and against inadvertent or unauthorized disclosure by the lawyer or other persons who are participating in the representation of the client.”(7) Utilizing generative AI may expose client information to both the developers of the AI tool and to third-parties, violating client confidentiality. 

An additional consideration for attorneys is whether utilizing AI to create documents falls under the scope of attorney work product. A recent federal case ruled that “ChatGPT prompts were queries crafted by counsel and contain counsel’s mental impressions and opinions about how to interrogate ChatGPT.”(8) This indicates that inputs from an attorney to an AI tool, because they contain an attorney’s mental impressions and opinion, may constitute a protected work product and be immune to discovery. However, courts have yet to fully address whether AI-generated outputs constitute opinion work product, ordinary work product, or something else entirely for the purposes of discovery.

Finally, the decision in Fortis Advisors, LLC v. Krafton, Inc. shows that attorneys will need to proactively inform clients of the ramifications of AI use.(9) Since clients may ask AI services questions even before litigation begins, the ruling in Fortis opens those queries up to discovery. Once litigation is reasonably anticipated, parties generally have a duty to preserve potentially relevant evidence. Clients may delete such seemingly innocuous conversations, or they may be deleted automatically by the AI service and thus expose clients to potential spoliation sanctions for the failure to preserve what is now discoverable evidence.

Conclusion

The novelty of widely-available AI systems means that many legal issues regarding the use of these tools are in flux. As Mr. Lee’s lurid ChatGPT searches indicate, careless use of AI for legal questions can have severe consequences. By now, most people have likely seen reports regarding AI “hallucinations” or other inaccuracies. But there are even more reasons to be wary when asking for help from AI services. People seeking legal advice should be cautious with what they ask AI tools, as even seemingly innocuous questions may ultimately become discoverable in litigation. Attorneys utilizing AI should opt for more secure products and avoid specific client information whenever possible. Options like Lexis’ Protégé and Westlaw’s CoCounsel both offer platforms designed with confidentiality protections intended for professional legal use.(10) However, even with these products, attorneys must review AI-generated content and consider client privacy whenever submitting information to third parties.

Takeaways:

• AI use in the legal industry has boomed over the past few years, and further advancements are expected to continue to increase AI use
• Clients seeking answers to their legal questions should recognize that their searches may be discoverable evidence. They should consult with attorneys before implementing AI inquiries for a specific legal situation and assume that any conversations with AI tools may later become discoverable and potentially be used against them.
• Attorneys utilizing AI should eliminate all client information from queries and meticulously verify the accuracy of AI responses before putting the responses to use
•  Both people seeking legal advice and attorneys should utilize AI tools with caution. In a shifting legal and technological landscape, the best practice is likely to assume that nothing shared with AI is private.
  

1 Guardian News and Media. (2026, March 10). Former Super Bowl champion asked Chatgpt about injuries before girlfriend’s death, Court hears. The Guardian. https://www.theguardian.com/sport/2026/mar/10/darron-lee-chatgpt-murder-court-hearing

2 Report on Artificial Intelligence in the Legal Profession, State Bar of Michigan https://www.michbar.org/AI (updated June 2025).

3 Id.

4 United States v. Heppner, 2026 LX 51521, at *7 (S.D.N.Y. Feb. 17, 2026)

5 Fortis Advisors, LLC v. Krafton, Inc., 354 A.3d 906 (Del. Ch. 2026)

6 Richter, M. (2025, October 24). See what legal professionals say about the role of AI and law. Thomson Reuters Law Blog. https://legal.thomsonreuters.com/blog/how-ai-is-transforming-the-legal-profession/#heading-2

7 2025 Model Rules of Professional Conduct Rule 1.

8 Tremblay v. OpenAI, Inc., No. 23-CV-03223-AMO, 2024 WL 3748003 (N.D. Cal. Aug. 8, 2024)

9 Fortis Advisors, LLC v. Krafton, Inc., 354 A.3d 906 (Del. Ch. 2026)

10 LexisNexis. (n.d.). Lexis+ AI Security information. Lexis+AI. https://www.lexisnexis.com/supp/lawschool/resources/faculty/lexis-ai-security-information.pdf.; Atwood, K. (2025, May 16). How in-house lawyers can use the power of AI on Westlaw Precision. Thomson Reuters Law Blog. https://legal.thomsonreuters.com/blog/how-in-house-lawyers-can-use-the-power-of-ai-on-westlaw-precision-with-cocounsel/