Your mid-month pulse on the signals shaping community & regional banking.
By Jim Perry, Senior Strategist
This month, several developments have me questioning assumptions that have shaped banking strategy for years.
Consumers are turning to AI platforms for financial guidance. Banks and technology companies are wrestling with how much authority AI agents should have over consumers' money. And despite strong financial results, new research suggests vulnerabilities within the banking system are increasing.
What interests me is the growing distance between what appears stable today and what may be changing underneath.
The midterm elections add another layer of uncertainty. With Democrats taking control of Congress, banking regulation, fintech charter approvals, and digital asset policy could face greater scrutiny in the year ahead.
Here are three signals worth your attention.
Signal #1 | The Battle for Financial Influence
What's happening
McKinsey's latest Global Banking Annual Review found that nearly one-quarter of surveyed consumers across 13 countries use generative AI for financial tasks. Among those using AI for banking activities, 74% turn to general-purpose platforms such as ChatGPT, Claude, or Gemini, roughly twice the share using their bank's chatbot.
Yet 62% would prefer AI-powered financial services and advice from their primary bank. Roughly half might turn to a third-party financial app if their bank doesn't offer comparable capabilities.
This month, Anthropic introduced personal finance capabilities in Claude, allowing consumers to connect financial accounts through Plaid, analyze transactions, and ask questions about their money. The service is currently read-only, so it cannot initiate transactions.
Why it matters
We've spent years teaching consumers to come to us. Think about how much time and money banks and credit unions have invested in their digital channels, all with the expectation that when people need something financial, they'll open our app. AI assistants are beginning to challenge that assumption. Consumers can now get answers to financial questions without ever visiting their institution's website or mobile app.
Influence doesn't require control of the transaction. A consumer can ask an AI assistant where to earn a better return on savings or whether they're paying too much for credit. The assistant doesn't need permission to move money to influence where it eventually goes. That's a competitive shift I don't think we're paying enough attention to.
Trust still gives financial institutions an opening, but it won't guarantee relevance. McKinsey's findings suggest consumers would prefer financial guidance from institutions they already trust. That's encouraging. But if we're not providing the help they're looking for, they'll increasingly find it somewhere else. We may be approaching a point where the institution holding the money isn't necessarily the one influencing what happens to it.
Questions for your leadership / board
How will our institution remain visible when AI assistants recommend financial products and providers?
What would make consumers prefer our financial guidance over that of a general-purpose AI platform?
Are we considering AI's impact on consumer behavior, or primarily its potential to improve internal efficiency?
Signal #2 | Who Controls the Agent?
What's happening
As AI assistants move from answering questions to completing transactions, financial institutions and technology providers are confronting questions about authority and accountability.
A Wall Street Journal report this morning describes how Stripe is building infrastructure for AI-initiated purchases, while JPMorgan Chase is restricting outside agents' access to customer accounts and transactions.
The OCC's November 4 supervisory framework emphasizes appropriate oversight of advanced AI, including human review and safeguards against fraud and cyber threats.
New research from Datos Insights, commissioned by BioCatch, provides the consumer perspective. Its survey of 3,000 digital banking consumers across six countries found:
53% selected real-time transaction alerts as an important control.
48% want approval requirements for larger transactions, and the same share want the ability to revoke agent access.
52% consider human review of challenged AI decisions an important trust factor.
56% believe their bank should bear responsibility for unauthorized or incorrect AI payments, regardless of how they occurred.
Perhaps most surprisingly, 51% would view their bank more favorably if it blocked AI agents from accessing their accounts.
Why it matters
Consumers are asking a different question than the industry. While banks and technology companies debate access, authority, and liability, consumers seem more concerned about something basic: Who's going to protect me if something goes wrong? The Datos research is a useful reminder that consumer confidence in agentic banking will depend on more than the technology's capabilities.
Our existing protections weren't designed for autonomous third parties. We've spent decades building systems to authenticate transactions, detect fraud, and resolve disputes. Now we're introducing AI agents that can act on someone's behalf, and it's not always clear how those protections will apply. Consumers want to know what an agent is doing, set limits on its authority, and be able to intervene when necessary.
Trust needs to become something consumers can actually experience. Community banks and credit unions talk a lot about trust as a competitive advantage. Here's an opportunity to demonstrate it. I don't think the answer is to block every outside agent indefinitely or allow unrestricted access. The next competitive advantage may belong to the institution that gives consumers the most control, not the AI agent the most freedom.
Questions for your leadership / board
Do we know which third-party AI tools can access customer or member financial information, and what those connections permit?
How would we authenticate and monitor transactions initiated by AI agents?
Can we clearly explain who is responsible when an agent initiates an unauthorized or unintended transaction?
Signal #3 | Strong Banks, Growing Vulnerabilities
What's happening
The OCC's Fall 2026 Semiannual Risk Perspective reports that profitability among OCC-supervised banks reached levels not seen in more than two decades during the first half of 2026.
For institutions under $30 billion in assets, net income increased 20% year over year, net interest margins improved by 19 basis points, and loan balances grew slightly more than 4%. Credit quality remained favorable.
But the OCC also warns that higher long-term interest rates are reducing securities valuations, competition for deposits could increase funding costs, and affordability pressures are affecting certain households.
Meanwhile, a November 12 analysis from the New York Fed found that three of its four measures of banking-system vulnerability are trending upward. Researchers identified growing exposure to liquidity stress, unstable funding, and potential forced asset sales, though their findings do not suggest an imminent banking crisis.
And then there's Nvidia. Jensen Huang's abrupt departure has raised questions about the financing arrangements supporting the AI infrastructure boom. The OCC had already flagged this area as an emerging concern, citing complex financing structures, rapidly depreciating technology, and interconnected borrowers.
Why it matters
I wouldn't confuse strong performance with resilience. Improving margins, healthy earnings, and favorable credit quality are good news, particularly after several challenging years. But a bank can be making plenty of money today while becoming more vulnerable to changes in funding costs, deposit behavior, or investment values. The New York Fed's research is a reminder that risks can be building even when the numbers we're accustomed to watching look pretty good.
Our assumptions about deposit stability deserve another look. Consumers already have more options for comparing rates and moving savings. Now consider what happens as AI assistants make those decisions easier and eventually act on them. AI isn't creating deposit mobility. It's potentially accelerating it by removing friction that has historically kept deposits in place. That doesn't mean agents are causing liquidity problems today, but it should prompt us to revisit how we think about deposit behavior.
Today's results don't tell us how prepared we are for tomorrow. The OCC's concerns about AI infrastructure financing offer another example of how new risks can develop outside traditional banking activities and eventually affect the broader financial system. For community banks and credit unions, the lesson is to look beyond current performance and consider how their institutions would hold up if the operating environment changed more quickly than expected.
Questions for your leadership / board
Do our liquidity stress tests account for faster deposit movement and renewed competition for funding?
Could strong earnings be obscuring emerging interest-rate exposures or concentrations?
Which assumptions about deposit stability and consumer behavior deserve reconsideration?
Closing Insight | The Assumptions Worth Challenging
As I look across these three developments, what stands out isn't that community banks and credit unions are suddenly at a disadvantage. It's that some of the things we've counted on for years may not be as dependable as they once were.
Consumers can trust their financial institution and still look elsewhere for advice. They can embrace AI-powered convenience while insisting on greater control. And banks can deliver excellent financial results while becoming more exposed to risks that haven't yet shown up in their earnings.
This is particularly relevant as many institutions head into strategic planning season. We spend a lot of time asking what we want to accomplish over the next several years. I wonder if we spend enough time challenging the assumptions behind those plans.
What are we counting on remaining true, and what happens if it doesn't? That question belongs on your next planning agenda.
P.S. A consumer warning sign worth watching: Reuters reports that U.S. banks collected $1.6 billion in overdraft fees during the third quarter, up 9.4% from the previous quarter and the highest quarterly total in four years. Economists suggest the increase could be an early indicator of growing household financial stress. I'd encourage community banks and credit unions to look beyond the revenue and ask whether more consumers are having trouble making ends meet—and what their institutions can do to help.
Quick ask: Which assumption deserves the most scrutiny at your institution right now: that consumers will continue looking to you for financial guidance, that existing controls can accommodate AI-initiated transactions, or that today's strong financial performance reflects lasting resilience?
Lots to consider this month. Comments always welcome.
