Think Twice Before Asking AI About Clients’ Estate Planning

We’ve all heard the saying: If you wouldn’t want it splashed across the front page of tomorrow’s newspaper (or The Daily Upside newsletter), it’s probably best not to say it at all.

The same rule applies when it comes to sharing sensitive estate planning information with general-purpose AI chatbots. It may seem like a harmless exercise to run one’s trust-funding strategy or family business succession plans past the likes of Claude or ChatGPT, but doing so can actually subject that information to future discovery if the plan is challenged in court, according to a team of specialist attorneys at ArentFox Schiff. So advisors and their clients should utilize extreme caution when utilizing public generative AI tools as part of the estate planning process, especially in situations where a lot of wealth is at stake or a future estate dispute seems likely. 

“It’s fraught with risk when we recklessly invite AI into the attorney-advisor-client relationship,” said Sarah Kerr Severson, a partner on ArentFox Schiff’s private wealth and tax planning team. “There’s no attorney-client privilege there. Courts have already confirmed that.” 

Risk Alert 

Back in February, a federal judge in New York issued a bench ruling holding that documents prepared using generative AI were not protected by attorney-client privilege or the work-product doctrine, a lesser-known legal rule that shields materials prepared in anticipation of litigation from discovery by an opposing party. The decision highlights the risk of using public-facing AI tools in legal matters, Kerr Severson said, particularly when such tools are used outside the direction of counsel and lack confidentiality protections traditionally afforded to attorney-client communications and other privileged materials.

“A client’s AI searches and documents do not constitute protected communications between attorney and client,” Kerr Severson said. “The privilege requires a trusting human relationship between a client and a licensed professional who owes fiduciary duties to the client and is subject to discipline if the duties are breached. Anything typed into AI chatbots should be assumed to be searchable, traceable and usable in a court of law.” 

That matters for a variety of reasons: 

• Clients’ private thoughts about their heirs, assets and estate mechanisms like trusts and wills could become public record. 
• Even if the exposure doesn’t result in adverse legal consequences, it could nonetheless be embarrassing or otherwise harmful to family relationships. 

Best Practices. This isn’t to say that generative AI shouldn’t play any role in estate planning whatsoever, Kerr Severson was quick to add. It can be very effective at educating clients about key concepts and vocabulary, for example. “The best practice is to keep things general,” she said. “Don’t use client names. Don’t use names of beneficiaries. Don’t put in a specific description of family assets or dollar amounts. Don’t use general AI models to create summaries or flow charts of the estate plan. Again, all of that could be discoverable.”