These days everyone and their dog is using Chat GPT, Claude, Gemini, Grok, and other Chat AI programs to look up legal cases and arguments. We have seen a number of legal AI hallucination fails with embarrassing and costly consequences to lawyers and self represented parties. MacLean Law dealt with Canada’s first legal hallucination case and warned of the dangers it presented to the legal system. Zhang v. Chen, 2024 BCSC 285 was MacLean Law’s winning landmark decision addressing the misuse of artificial intelligence (“AI”) tools in a British Columbia family law case.
We warned if lawyers and parties were not diligent in having a human in the loop the Canadian legal system would face an existential threat. Sadly, since then Canada had had dozens of similar cases. However, the pandora’s box of legal AI research also contains other traps for the unwary. Today Lorne MacLean KC and Fraser MacLean provide more thought leadership on the foibles of careless use of legal AI and public and enterprise level artificial legal intelligence. There are real dangers to using AI Chatbots in ultra high net worth (UHNW) family law cases. As the cases below emphasize unsupervised or informal use in sensitive legal contexts can:
- undermine privilege,
- endanger or expose trial strategy,
- undermine settlement options, and
- potentially create reputational exposure.
How To Stay Safe
Rule # 1: If you are represented by a family lawyer in your case, do not use AI regarding any aspect of your case at all.
Rule # 2 If you are an owner of a firm or a lawyer understand the dangers of “Shadow AI” which is the use of personal or unauthorized AI accounts for firm or client work.
- Shadow AI is the single greatest AI-related confidentiality risk facing law firms.
- Any use of personal AI accounts for client work may result in client data being subject to consumer-grade data terms including potential model training.
- Shadow AI use by any firm personnel must be strictly prohibited and treated as a serious disciplinary matter.
MacLean Law handles ultra high net worth international family law cases and has set numerous precedents and records in the family law and we wrote this article to help family law clients and self represented parties to stay safe.
Loss of Privilege For AI Legal Research
MacLean Law handles ultra high net worth and international family law cases where there is no margin for error. Legal clients are now more than ever resorting to potentially catastrophic unsecure AI platforms. MacLean Law wants to keep you from losing potentially millions and parenting time.
A novel recent New York USA decision in US V Heppner held that research and chats by a client with a publicly available, non-enterprise, generative AI platform, (Harvey, Alexi etc.) THAT WERE NOT made at the request of counsel, are not protected by attorney-client privilege or the work product doctrine nor the litigation work product exception.
Heppner held that AI platforms are not lawyers, nor are they confidential (even when consumers use a paid account). The AI platforms’ terms of service allow for data review and third-party disclosure.
Litigation Work Product and AI Legal Research Tel: 604 602 9000
But wait in breaking news Warner v. Gilbarco, Inc another Michigan USA case held that it was protected as a “litigation work product” by concluding AI is a tool, not a person. On this approach, using an AI platform isn’t disclosure to a third party in the way that matters for waiver — because work product waiver requires disclosure to an adversary, not merely to any third party. The Court dismissed the application to produce the AI legal research searches.
“In the end, both sides of this dispute seek to obtain each other’s thought processes, while shielding their opponent from discovery of their own. The Court will uphold the protections afforded the thought processes and litigation strategies of both sides and will order production of neither.”
Dangers Of Using AI Chatbots For Research in UHNW Family Cases
Who is right? What will Canada do? Stay tuned and stay safe on the unrepresented litigant front!
Access To Justice And Legal AI
QUERY? If an unrepresented client’s legal research is not privileged while a lawyer’s research is, are we depriving self represented litigants access to justice? Will they be afraid to do research and as a result lose in court? We like the ‘litigation work process” analysis better so there is a level playing field.
Is the situation different if the lawyer uses an enterprise grade program like Harvey or Alexi?
Maybe, but privilege still relies on the lawyer’s conduct. If a Canadian lawyer uses Alexi or Harvey or some other enterprise program but then copies and pastes that confidential research into a public forum, or shares their Harvey login with someone outside the firm, the privilege can still be lost. The tool is a “secure repository of data and research” but the lawyer must keep the lid closed. Make sure your contract has data isolation to your firm, that the program doesn’t retain it and that your firm maintains rigid access controls.
What about free legal chatbots and programs created by some jurisdictions to facilitate access to justice in Canada?
Some jurisdictions are now allowing legal chatbots to help assist self-represented parties. However, the danger with these no matter how well intentioned is the following:
A chat with an AI chatbot or a public forum, even if intended to help the public (access to justice), does not create a fiduciary, confidential relationship with a licensed professional.
TAKEAWAY: Tel: 604 602 9000
Lawyers should add a section to their retainer agreement warning clients not to input any specific facts and more specifically any specific legal advice you have provided to them about their case to avoid losing solicitor client privilege. Lawyers should make sure any program they use is secure and does not share data outside of the firm, that no data is retained in the program that is not limited to the firm and that access to the secure digital file cabinet is secure.
Add potential loss of litigation privilege to to your legal AI fear index!
