Like any piece of technology, AI has a learning curve. Although anyone can type a prompt into a ChatGPT or Claude chatbox, it often takes repeated rounds of prompts (and sometimes extensive back-and-forth) to actually get what you want out of it. As a result, there are some people who try it out, play around for a little while, and eventually lose interest, while others will keep at it to learn the art of prompting and eventually move on to more complicated uses like analyzing documents or creating agents to perform routine, multi-step tasks. But even though there's a degree of skill involved in prompting, it's still done in plain English: Building something with AI doesn't require learning a whole other language (as it does if you want to learn how to build things in, for instance, Python or Java). Which means that it doesn't take much more than a weekend of work to get pretty good at getting AI to do at least moderately complex tasks, including pulling data from other sources (such as off the Internet or from files on the user's computer) to perform tasks with.
On the one hand, this might seem like a positive for advisory firms, since with a little creativity and a relatively minimal amount of one-time effort an employee can revamp a manual process to be automated by AI, saving them time and allowing them to be more productive elsewhere. But on the other hand, there are fairly significant downsides to advisory firm employees having the ability to build AI tools from scratch, most prominently when it comes to client data (which advisory firms are required to protect under SEC Regulation S-P). If the employee is using a 'free' version of ChatGPT or Claude, any information (including client data) that they put into the AI will generally, by default, be used by the AI model for training purposes, effectively feeding the client's personal information into the AI model. And even if an employee takes care not to directly enter client information into the AI (or uses a paid version that doesn't train on information entered by the user), an AI agent that they create to perform a certain task could still find its way around whatever safeguards the RIA has in place to protect client data. Having a handful of (amateur) software developers on staff creates risk for advisory firms that often take great pains to evaluate third-party vendors precisely to manage the risk of client data being inadvertently exposed.
It's notable, then, that the SEC is reportedly taking an interest in how advisors train and supervise their employees regarding their use of technology, including AI. Which reflects the inclusion of "training and security controls that firms are employing to identify and mitigate new risks associated with [AI]" on the SEC's list of 2026 exam priorities, and their particular focus on compliance with Regulation S-P (which took full effect in June of 2026).
Notably, the SEC hasn't announced whether it has fined or otherwise warned or disciplined any advisory firms regarding their employees' AI use. On one level, it would seem to be similar situation in recent years when the SEC levied fines on multiple firms for failing to supervise employees' use of off-channel communications (such as sending messages about client work to colleagues via text or messaging app rather than going through company-approved and monitored channels). Every company that has employees who act as 'citizen developers' using unapproved AI tools to ask client-related questions or build automations for client work could hypothetically find itself at risk of a similar fine – however, it's unclear as of yet whether the Trump-era SEC will be as willing to levy such fines on RIAs as it was under the Biden administration. It's possible that, rather than starting with outright fines, the SEC takes what it learns from this round of examinations and issues new guidance to help RIAs clarify their own policies around managing and supervising AI at their firms.
In the meantime, though, it's important for RIAs to be aware of the risks that AI tools – both authorized and unauthorized – present, and account for both in their policies and procedures. An RIA might decide to simply ban all use of Claude or other general-use AI apps, but that won't always mean that employees will abide by those restrictions – it may simply push their use to where the RIA can't see it. It may be better instead for RIA leaders to first take stock of how their employees want to (or do) use AI, which tools they use, and whether there's a way to use them within the firm's client data protection policies. That way, they can vet the tools through the firm's standard approval process (possibly signing up for an enterprise license where the data is auditable and not used to train the LLM), and supervise how employees use them (including archiving the prompts that the employees use, which may be fair game for the SEC to request).
The bottom line is that, while it may be a stretch to say that AI use by employees is an inevitability, there's at least a strong likelihood that advisory teams will have at least one or two people who are interested enough in the widely-available AI tools to make some use of them for work purposes. Firms that don't yet have a policy on their employees' AI use, or who have a policy (or even an outright ban on AI tools) that employees aren't following, might find themselves needing to make adjustments sooner than later if the SEC continues to probe the topic in its examinations – even if it isn't yet handing out outright fines or disciplinary measures.
This article first appeared on the Nerd’s Eye View at Kitces.com at https://kitc.es/advisortech-oct2026, and has been reprinted here with permission.
Ben Henry-Moreland
Ben Henry-Moreland is a Senior Financial Planning Nerd at Kitces.com, where he specializes in writing and speaking on financial planning topics including tax, practice management, and technology. He also co-authors the monthly Kitces #AdvisorTech column. Drawing from his experience as a financial planner and a solo advisory firm owner, Ben is passionate about fulfilling the site’s mission of making financial advicers better and more successful.
Michael Kitces
Michael Kitces is Head of Planning Strategy at Focus Partners Wealth, which provides an evidence-based approach to private wealth management for near- and current retirees, and Focus Partners Advisor Solutions, a turnkey wealth management services provider supporting thousands of independent financial advisors through the scaling phase of growth.
In addition, he is a co-founder of the XY Planning Network, AdvicePay, fpPathfinder, and New Planner Recruiting, the former Practitioner Editor of the Journal of Financial Planning, the host of the Financial Advisor Success podcast, and the publisher of the popular financial planning industry blog Nerd’s Eye View through his website Kitces.com, dedicated to advancing knowledge in financial planning. In 2010, Michael was recognized with one of the FPA’s “Heart of Financial Planning” awards for his dedication and work in advancing the profession.
Kate Maier of Wealth Enhancement explains why UHNW clients aren’t waiting until death to give.
Bank of America data shows near-term financial pressure mounting even as average balances gain nearly 10% since year-end.
Wells Fargo adds two bank-channel recruits, UBS lands a JPMorgan Chase veteran, and a multi-family office lifts Atlas past $1B
“The state comes down on you in years when you make 40% returns for clients,” said Bill Swayne, the advisor and son of Candace Swayne.
The Independent broker-dealer giant has had “active conversations” with the Claude maker, which is days away from its expected IPO
Growing wealth, evolving priorities, and untapped opportunities are reshaping how women approach long-term investing.
Knighthead Life entered the market with a competitive MYGA. A strong launch earned advisor confidence and paved the way for FIAs.