TL;DR
I have been in some version of the same conversation about AI inside credit unions for the last twelve months. The conversation starts well. Somewhere in the middle, one of three objections shows up. The conversation stops moving forward. Below are the three objections, why each of them is reasonable, and the framing I have seen actually move the room past them. None of this is about overpowering the skeptic. It is about taking the objection seriously enough to answer it.
Credit unions are not slow because credit union leaders are unsophisticated. They are slow because the people responsible for risk inside a credit union are doing their jobs. The objections I am about to walk through are the objections of people taking their jobs seriously. The reason the conversation stalls is not the objection. The reason the conversation stalls is that the room does not have a shared way of working through the objection.
Here are the three I keep hearing, in roughly the order they arrive in a meeting.
Objection one: “Our members do not use AI yet”
This one is the most common and the most quietly wrong.
The objection assumes that AI use is a category, like online banking, that members either have or do not have. The data on AI use does not support that view. AI tools are showing up in member behavior in ways that do not look like “using AI.” A member running a rate comparison inside ChatGPT does not think of themselves as an AI user. A member asking Perplexity which credit union has the highest savings rate in their county does not think of themselves as an AI user. A member who reads a Google AI overview before clicking through to your site does not think of themselves as an AI user.
The framing I have seen work is to separate AI tool usage from AI surface exposure. Your members are not necessarily using AI tools the way you use them. They are absolutely exposed to AI surfaces in their financial decision making, because the surfaces are now embedded everywhere they search.
The right question is not “are our members using AI.” The right question is “are AI tools mentioning us when our members are making decisions.” That question has an answer, and the answer is measurable.
Objection two: “Our regulator is not ready”
This one is the most reasonable and the most often misused.
The regulators are paying close attention to AI inside financial institutions. The NCUA, the OCC, the FDIC, and the state regulators have all issued guidance, examination expectations, or model risk frameworks in the last eighteen months. The pace of guidance has been quick, the substance of it has been measured, and the direction has been consistent. AI is permissible. It is also subject to model risk management, vendor management, fair lending review, and consumer protection oversight.
Where this objection gets misused is in the leap from “the regulators are paying attention” to “the regulators will not let us.” The first is true. The second is not. The regulators expect you to govern your AI use. They do not expect you to avoid it.
The framing I have seen work is to separate the production decision from the governance decision. The governance decision is real and unavoidable. Model risk management. Vendor management. Documentation. Bias testing. Consumer protection review. These are the work, and they are the work whether the credit union is using AI for member emails, for credit decisioning, or for back office automation.
The production decision is whether to actually deploy. That decision can be made one use case at a time. Each use case carries its own risk profile, its own governance requirement, and its own documentation burden. Treating “AI” as one thing makes the governance impossible. Treating each use case as its own decision makes it tractable.
Objection three: “Our members will not trust it”
This one is the most emotional and the hardest to engage in the meeting.
The objection comes from a real place. Credit unions earn their member relationships. Trust is the asset on the balance sheet that does not show up. The fear that AI will erode the trust is the fear of damaging the only thing that makes a credit union different from a bank.
I take this one seriously every time. The framing that works here is not about overcoming the objection. It is about respecting the underlying value while showing the room where the actual trust risk lives.
The trust risk is not in AI itself. The trust risk is in AI that is invisible to the member, that produces bad outputs that the member cannot trace, and that the credit union cannot explain. The trust risk in a transparent AI system is roughly the same as the trust risk in any other technology the credit union has adopted in the last twenty years. The trust risk in an opaque AI system is high.
The conclusion the room can reach together is this. The way to protect member trust in the AI era is to deploy AI in ways the member can see, understand, opt out of, and trust. That is a design constraint. It is not a reason to avoid the decision.
What changes the room
The three objections share a structure. Each one is a reasonable concern dressed up as a stopping point. The work of moving the room is not to dismiss the concern. The work is to separate the concern from the stopping point.
Members and exposure are different from members and tool usage. Governance is different from prohibition. Trust risk is different from technology risk. Each separation gives the room a way to keep moving without telling anyone their concern was wrong.
The marketing leader in the room is often the person best positioned to do this work, because the marketing leader is fluent in both the member language and the institutional language. The CIO can do it. The CRO can do it. The CMO often has the lightest political tax for trying.
The bottom line
The AI conversation inside credit unions is not stalling because credit unions are slow. It is stalling because the room does not yet have a shared way of answering three reasonable objections. The framings above are not magic. They are working. The credit unions where the conversation has moved past these objections are starting to ship.
The credit unions where the objections still own the room are losing months they will not get back. Twelve months of strategic ground is meaningful in a year when the AI surface is reshaping discovery faster than any other channel.
Take the objections seriously. Then move.
Kevin Farley writes about AI visibility, AI readiness, and strategic growth for financial services. Read more on the blog.