How Banks Are Using AI in the Boardroom

The use of artificial intelligence (AI) is moving beyond bank operations and into the boardroom. Bank Director’s soon-to-be released 2026 Governance Best Practices Survey, sponsored by Bradley Arant Boult Cummings, shows that over 23% of all banks say they use AI or automation to transcribe board meeting discussions, prepare meeting minutes or compile materials for the board book.

Sonata Bank, a $283 million institution based in Brentwood, Tennessee, which specializes in banking franchise restaurants and providing sponsor banking services to fintechs, is using enterprise-level AI tools for all of the above. Chief Innovation Officer Will Rhoads says the bank is deploying secure solutions from several providers. “For all of our [board] committees, we use [Microsoft] Copilot to prepare meeting minutes,” he says. “Somebody was taking the time to do that. Now, it’s all done by Copilot. Those are then produced and sent directly to the board.”

Sonata Bank CEO Wendell Bontrager says the board meetings are also recorded and then transcribed by AI. “We still have to review those notes because, right now, it still has not yet been trained enough over the months to hone in on how to convert that into minutes that we can just simply quickly review and check,” he says. “It gives too much detail, and so, we have to weed out the details to make sure it’s appropriate for the audience.”

The $6.3 billion United Fidelity Bank, based in Evansville, Indiana, is also using AI to transcribe board discussions. Paul Becker, chair of the board’s technology committee, calls it a major time saver. “At the board level, which can be a five-hour meeting, it used to take two days for our board secretary to prepare the minutes. It’s now four hours,” he says.

Sonata Bank is also utilizing agentic AI to help its board stay up to date on a number of issues. Through the use of Anthropic’s Claude Managed Agents, Rhoads’ team is able to direct AI agents to comb through publicly available information on a variety of topics, and then create a summary of its findings. One of those topics is on the ways AI is being used in financial services. Bontrager sends the agent’s summary in a weekly email to his directors. “The starting point is bringing the board up to speed so everybody knows what AI is fundamentally,” Bontrager says. “But it’s really applying that directly to our industry and then — how are we looking to utilize AI solutions? How are we putting guardrails around it? All that framework we’re just getting started on.”

Sonata Bank AI agents are also pouring through new banking legislation and regulation for another weekly email Bontrager is starting to send out to the board to keep them abreast of policy developments. Rhoads says there are also agents looking at compliance issues, fraud and cybersecurity threats as well. “We then surface relevant items back to the board that are being pulled out of that,” he says. “Because even though that is surfacing more relevant items, it’s still a lot.”

Since all of those agentic models are probabilistic, meaning they produce some level of uncertainty and randomness, Rhoads says a human still reviews all of their outputs in accordance with Sonata Bank’s AI acceptable use policy.

Managing Risk and Governance
While Sonata Bank has an established AI use policy, Robert Maddox, a partner at Bradley who provides legal representation to financial institutions, says that is likely not the case for every community bank using AI in the boardroom. “Adoption is running ahead of governance,” he says. “So, people are using it and they don’t have a tool to govern it.”

Governance in relation to board discussions and materials are crucial, because they often contain highly confidential information. “I mean, you’ve got strategy, you’ve got M&A, credit concentrations, you can have exam findings,” Maddox says. “And the second any of that moves through a ChatGPT or a Claude or a Harvey or whatever you want to use, you’re putting your information in someone else’s model.”

Maddox says banks must use a paid, secure enterprise version of those third-party AI platforms to have control over what information leaves the bank. Many banks are also using online board portals provided by companies like Diligent and OnBoard Meetings, which provide secure, digital platforms for scheduling and hosting meetings, voting, sharing sensitive documents and more. These companies claim their portals offer built-in governance, risk and compliance capabilities, all of which apply to the AI tools that are continually added to them.

Brian Stafford, CEO of Diligent, says financial institutions are demanding more AI tools than ever before. “What’s changed over the last few years is that the conversation has expanded beyond digitizing the board book,” he says. “Banks still need a secure portal, of course, but increasingly they also want tools that help directors absorb information faster, connect issues across materials and focus on the questions that matter most. That is where AI starts to become very relevant.”

But Stafford and Maddox both say humans must review any work the AI tools are producing. “In banking especially, AI has to operate inside governance controls,” Stafford says. “That means strong permissions, encryption, audit trails, clear separation between draft outputs and final approved records, and human oversight before anything is treated as part of the formal board file.”

For Maddox, human oversight is especially crucial when it comes to any financial data produced by AI. That’s because generative AI can make up information rather than simply reading and reporting. “The numbers could be completely wrong,” he says.

What Regulations Should You Follow?
Regulators have yet to release guidelines on anything related to the use of AI in the boardroom. But Bontrager says in a recent meeting he had with his bank’s regional director for the Federal Deposit Insurance Corp., he asked what the FDIC’s stance was on the use of AI inside the bank. “And his answer was, ‘You don’t need to get our permission to use it. Just keep us posted on how you’re using it, and make sure you put the right risk framework around it,’” he says.

Bontrager’s takeaway was that Sonata Bank needs to approach the use of AI in the boardroom the same way it approaches its loan policy, wire policy, ACH policy and any other policy that relates to personally identifiable information (PII). That’s why the use case in the boardroom doesn’t include client data. “If you think about us complying with current regulations and putting the guardrails around client data and information, we’re applying the same principles in how we apply guardrails to AI.”

Maddox says that until regulators put out specific guidelines on AI use, banks would be wise to follow the Interagency Guidance on Third-Party Relationships: Risk Management from the FDIC, Office of the Comptroller of the Currency and the Federal Reserve. “It sets an expectation of genuine diligence on vendors overall,” he says. “And so what we would say is that fits directly into that key third-party risk management piece.”

Maddox also encourages banks to write down what established guidance they are following for AI use, and exactly how they are following it. “Write it now and send it to your regulator and say, ‘This is what we’re thinking,’” he says. “Either they’re going to write it for you or you can write it for yourself. You’ve got to control your own narrative. I think that’s important.”