AI financial advice is becoming a second opinion for wealthy clients

AI financial advice is no longer a fringe habit for wealthy investors with access to professional money managers. Leaders at wealth firms told CNBC that clients are asking tools such as OpenAI’s ChatGPT and Anthropic’s Claude to review portfolios, tax ideas and advisor recommendations, creating a new pressure point for an industry built on trust.

Matthew Fleissig, CEO and co-founder of Pathstone, told CNBC he believes ChatGPT may already be “the single largest investment advisor in the world.” Pathstone is a registered investment advisory firm with $185 billion in assets, according to CNBC.

The shift does not mean high-net-worth clients are dropping their advisors. Executives described AI as a second set of eyes that can help clients ask sharper questions. A large language model, or LLM, is an AI system trained to generate text-based answers from patterns in data, but it can also produce confident-sounding mistakes.

Can AI replace a financial advisor?

Wealth managers interviewed by CNBC said AI can help investors compare ideas and prepare for meetings, but it still has limits as actual advice. Fleissig said chatbots can run strong analysis, yet they do not replace the judgment, relationships and access that human advisors bring to complex financial decisions.

Vince Lumia, who oversees more than 16,000 financial advisors at Morgan Stanley, told CNBC that the human side of advice tends to look less valuable when markets are rising. He said clients often want more guidance when volatility or uncertainty returns.

Morningstar’s Sean Dunlop told CNBC that AI is unlikely to eliminate traditional wealth managers, but it could change the economics of the business. He said AI tools may improve service, allow fewer advisors to oversee a larger pool of assets and encourage some customers to manage money on their own. Dunlop also said wealth management stocks have already weakened as AI tools have moved into parts of the industry, including tax planning and personal finance.

How clients are using chatbots

Pamela Lucina, Northern Trust’s chief fiduciary officer and head of trust and advisory practice, told CNBC she began seeing clients check the firm’s advice with AI roughly 18 months ago. She said the behavior has become more common and often pushes conversations away from basic information gathering and toward the client’s goals.

Lucina said prospective clients are also using LLMs to prepare detailed requests before deciding whether to hire Northern Trust. She estimated that about half of clients now send formal requests for proposals, including some with around $100 million in assets, a process she said used to be more typical among billionaire clients.

Michael Zeuner, managing partner at WE Family Offices, told CNBC his firm noticed clients openly running portfolio recommendations through LLMs about five months ago. He said the practice is now common enough that the firm is considering whether to test every recommendation in Microsoft Copilot so advisors can anticipate client questions.

Where AI advice can go wrong

Executives also flagged risks. Zeuner told CNBC that a client who uploads a trust document and asks about it later may get an answer with invented details. He also cited a case where ChatGPT told a client two ETFs, or exchange-traded funds, looked identical even though one tracked an index by equal weight and the other by market capitalization weight.

Lucina said that when she asked an LLM for examples of capital gains tax savings, the math was wrong. She told CNBC she finds AI useful for looking for weaknesses in an argument, but said it is often wrong on advice.

Privacy is another concern. Zeuner said financial firms often use enterprise AI plans with stronger data protections, including WE Family Offices’ agreement with Microsoft Copilot that firm data will not train public models. He said clients using personal AI accounts should understand where meeting transcripts, financial details or identifying information are stored and who can access them.

This story draws on original reporting from CNBC.