Lawyers warn of serious risks in using AI as wealth advisory
Artificial intelligence has become all the rage among high-net-worth clients looking for quick answers to complex legal questions. And honestly, it makes sense, right? After all, it is way easier to open ChatGPT than to schedule a meeting with a specialized attorney, wait days for a response, and still get a bill at the end.
The problem is that this convenience comes with a price tag that can end up being far more expensive than any legal fee.
Consider this real-life situation: a high-net-worth client in Florida consulted his attorney, Tasha Dickinson, a partner at Day Pitney, excited about a tax strategy that AI had suggested. The idea involved creating a community property trust, an interesting option for married couples. But there was one small yet devastating detail: his wife had recently passed away. That type of trust requires two living spouses. The AI simply did not know that. Or rather, it knew the strategy, but it knew nothing about that client’s actual life.
According to Dickinson herself, when she pointed out the issue, there was silence on the other end of the line. The client’s response was something along the lines of: the AI thought it was a good strategy. And the attorney summed it up perfectly: maybe in the universe of generic possibilities it is a good strategy, but it was definitely not a good strategy for that specific case.
And that is exactly where the danger lies 👇
It is not that artificial intelligence is useless for legal matters, far from it. But attorneys who specialize in high-net-worth clients are raising serious red flags about the risks of using these tools to make real decisions involving trusts, taxes, estate planning, and asset protection.
Clients are using AI to second-guess their own lawyers
Dickinson says she gets calls every week from clients asking about legal advice they got from ChatGPT, Claude, or other AI chatbots. Some do not even admit the source was AI, but she can tell from the line of questioning they present.
And she is not the only one noticing this trend. Robert Strauss, a partner at Weinstock Manion, reported that several clients have been uploading trust documents into AI systems and coming back with lists of questions and editing suggestions. This forces the attorney to defend their own work and explain why the AI recommendations do not apply to that specific situation.
According to Strauss, the questions themselves are not the problem. What happens is that the process ends up consuming far more time than would normally be needed. He estimates that his team spends two, three, or even four extra hours dealing with issues that, so far, have not produced a single practical contribution. In his words, he has yet to receive one viable suggestion from this process.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
The most concerning outcome, according to Strauss, is the erosion of trust between client and attorney. When AI contradicts the professional, even when it is wrong, it creates unnecessary tension that can compromise the entire advisory relationship.
What AI actually knows and what it completely ignores
Language models like ChatGPT, Claude, Gemini, and others are trained on massive volumes of text, including legislation, legal doctrines, case law, and specialized articles. This means that, in theory, they can explain complex concepts in tax law, estate law, or asset management in a way that is quite accessible and, often, accurate. For introductory research or to understand the vocabulary before a meeting with your attorney, these tools can be genuinely useful.
The problem starts when the client moves from research to decision-making.
Artificial intelligence does not have access to the client’s life history. It does not know that a spouse passed away, that there was a recent change in tax domicile, that there is ongoing litigation among heirs, or that a particular asset was informally transferred years ago. It responds based on what was asked, not based on what needed to be considered. This creates a silent trap: the answer looks complete, coherent, and confident, but it is built on a hypothetical scenario that does not match that specific client’s reality.
Ed Renn, from the firm Withers, gave a striking example. A client wanted to transfer unlimited assets to his wife after receiving that advice from ChatGPT. The detail the client did not mention to the AI? His wife was foreign-born, which means he could not take advantage of the unlimited marital deduction without a special type of trust. As Renn put it: if you do not know exactly what you are doing, it is garbage in, garbage out.
Renn added that AI tools seem to make more mistakes when the subject gets more complex, such as international taxation, and that they frequently are not up to date with new legislation or IRS guidance.
The false sense of knowledge
Dickinson pointed out something very relevant about the profile of clients who use AI the most for legal matters. According to her, many are successful entrepreneurs, smart people with a natural drive for seeking knowledge. And it is precisely this profile that leads them to believe they understand more about the subject than they actually do.
Using these AI tools, according to the attorney, creates a false sense of knowledge. The friendly interface, the articulate responses, and the speed of delivery create the impression that the user is receiving personalized and reliable advice, when in reality they are getting generic answers based on statistical text patterns.
In a sense, as Dickinson herself acknowledged, this is not exactly a new problem. Clients have always brought suggestions to their attorneys based on conversations with friends at the golf club or articles they read. The difference is that AI represents, in her words, a more evolved form of cocktail party conversation. The tech veneer makes the suggestion seem far more well-founded than it actually is.
Dan Griffith, director of wealth strategy at Huntington Bank, brought another important perspective. According to him, deciding how to transfer wealth to your loved ones requires a much more complex discussion than ChatGPT is prepared to facilitate. There are rarely simple answers when deciding, for example, how to divide assets between children from a first marriage and a second spouse.
Griffith gave an example: if a client asks whether a certain trust will guarantee that their son gets access to the funds at some point, the answer should not simply be yes or no. The answer should be: tell me more about your relationship with your son, or what is the family situation. According to him, AI tends to be overly solution-oriented, trying to find some way to get to yes, without doing a good job of investigating what is really behind the question.
Data privacy: the risk nobody is seeing
Beyond the legal errors themselves, there is another problem growing quietly that deserves full attention: data privacy and the breach of attorney-client privilege. When a client types a legal question into an AI platform, they frequently include extremely sensitive information. Asset values, portfolio structures, beneficiary names, details about litigation, health information that impacts estate planning, data about businesses and partners. All of this is entered into systems that have complex privacy policies and that may use this data to train future models.
Strauss revealed that his firm is currently revising client contracts to include a specific warning: the use of AI chatbots to discuss legal strategies may void attorney-client privilege. This is not an exaggeration or speculation. In February, a federal judge in the United States ruled that a criminal defendant’s conversations with Claude about his legal defense strategy were not protected by attorney-client privilege.
Dickinson herself confessed that this is the point that keeps her up at night. According to the attorney, what concerns her is not the fact that AI gets things wrong sometimes, because she can correct those errors. It is also not the fact that clients check her work using AI, because she has a lot of confidence in what she does. What truly concerns her is that by uploading documents and running these searches through AI, clients are waiving attorney-client privilege, and that is a massive problem.
Griffith added a practical warning: asking a chatbot how to protect your assets with a prenuptial agreement, or how to sell a business while paying fewer taxes, for example, could be used as evidence against the client in court. It is a cruel irony, but seeking savings and convenience can result in exactly the opposite.
Why high-net-worth clients are the most vulnerable
It might seem counterintuitive, but the wealthiest individuals tend to be exactly the ones who take the most risk when using AI for legal advice without specialized oversight. Part of this has to do with the profile of these people: they are executives, entrepreneurs, and investors accustomed to making quick decisions, trusting data, and optimizing processes. When an AI tool offers a detailed answer in seconds, the natural instinct is to treat it as a solid starting point or even as a definitive solution.
Another important factor is the complexity of the wealth structures involved. High-net-worth clients typically operate with a combination of trusts, holding companies, international accounts, investments in multiple countries, and estate plans that took years to build with the help of entire teams of attorneys and accountants. Consulting an AI about one isolated piece of that puzzle without considering the systemic impact is like asking someone about a single gear without showing them the whole machine. The answer might be technically correct for that gear, but completely wrong for that machine.
And even though high-net-worth clients have the access and financial ability to hire the best legal professionals available, as Griffith observed, they, just like everyone else, enjoy the convenience that AI offers. Dickinson added that the financial savings are also attractive. Paying for professional services is never anyone’s idea of a good time, she said. The combination of convenience, perceived savings, and entrepreneurial self-confidence creates the perfect scenario for risky decisions.
Lawyers make mistakes with AI too
It is worth noting that the trend is not one-sided. Many attorneys also use AI in their professional and personal lives. And this has already led to embarrassing and widely reported situations, such as court filings that included completely fictitious citations generated by artificial intelligence. These cases reinforce that the tool, without proper oversight, can fool even trained professionals.
However, as the attorneys interviewed pointed out, few clients are familiar enough with both AI and the law to write an effective prompt. The quality of the answer depends directly on the quality of the question, and formulating the right question requires prior knowledge that most people simply do not have when the subject is high-complexity estate and tax planning.
The irreplaceable role of specialized legal counsel
None of this means that technology should be thrown out. Quite the opposite. Top-tier legal advisory firms are already incorporating AI tools into their own workflows, but in a completely different way: with specialized human oversight, with anonymized data, with careful review of every AI-generated response, and with full awareness of these tools’ limitations. This approach allows them to leverage AI’s speed and analytical capacity without sacrificing the human judgment that complex situations demand.
What the experts emphasize repeatedly is that the value of legal counsel for high-net-worth clients is not just in the attorney’s technical knowledge, but in the relationship built over time. An attorney who deeply understands the client’s situation, their family, their long-term goals, their fears, and their priorities can identify problems that no AI would ever catch, simply because those problems were never typed into any text box. The Florida case is a perfect example: the attorney knew the wife had passed away. The AI had no way of knowing.
Structures like irrevocable trusts, international estate plans, and asset protection strategies require a holistic view that integrates the legal side, the tax side, the family side, and the emotional side of every decision. AI tools are excellent for expanding knowledge, speeding up research, and democratizing access to information that used to be restricted to those who could afford hours of consulting. But when the subject is the real protection of a lifetime’s worth of wealth, specialized human judgment remains, by far, the most valuable asset a client can have on their side. 💼
As Dickinson herself summed it up perfectly: AI can offer a good strategy in the generic universe, but be a terrible strategy for you. And at the end of the day, it is precisely that for you that makes all the difference between advice that protects and advice that puts everything at risk.
