AI only works for credit unions when advocacy, compliance, and member-centric design work together. Here’s a practical playbook to do that in 2025.
Advocacy, Compliance & AI: A New Playbook for CUs
Most credit union leaders I talk to feel the same tension: members want faster digital experiences, regulators want tighter controls, and staff are already stretched thin. Something has to give.
Dan Berger, President and CEO of NAFCU, summed up the real goal nicely:
“We want to create a legislative and regulatory environment so credit unions don’t just survive, but thrive and grow.”
Thriving in 2025 isn’t just about surviving new rules or adopting another shiny tool. It’s about using AI for credit unions in a way that respects compliance, strengthens advocacy positions, and keeps member-centric banking at the center of every decision.
This post connects three pieces that usually live in separate conversations: advocacy, compliance, and AI. If you’re responsible for strategy, risk, or member experience, this is the lens you need.
Why Advocacy & Compliance Matter Even More in an AI Era
Advocacy and compliance used to feel like back-office functions. Today, they directly shape what you can build with AI, how fast you can launch it, and how much value your members actually see.
Regulation is now a product design constraint
When Dan Berger talks about NAFCU’s mission—advocacy, compliance assistance, education—he’s talking about the foundations that make innovation possible. For AI projects, those foundations are non‑negotiable:
- Data usage rules define what training data you can use for models
- Fair lending and UDAAP shape how AI can support loan decisioning
- Privacy laws govern how you store, analyze, and share member data
- Cyber and fraud rules affect how you design AI-driven fraud detection
If your AI roadmap ignores these, you don’t have a roadmap—you have a risk exposure plan.
Advocacy determines the AI playing field
Trade groups like NAFCU are at the table when policy makers:
- Debate AI use in underwriting
- Consider new requirements for model transparency
- Weigh in on data privacy and consumer consent
The clearer you are about how you want to use AI for member-centric banking, the more effectively advocacy teams can argue for practical, innovation‑friendly regulation instead of blunt, restrictive rules.
This is why I push leaders to stop treating advocacy as “someone else’s job.” Your AI strategy and your advocacy posture are now tightly linked.
Turning Compliance from AI Roadblock into AI Design Partner
Here’s the thing about AI in credit unions: most projects stall not because the tech doesn’t work, but because compliance sees it too late.
Bring compliance in at the whiteboard, not at launch
Traditional pattern:
- Business team defines a cool AI use case
- Vendor pitches a solution
- IT runs a proof of concept
- Compliance sees it in month 6 and pulls the emergency brake
A better pattern for AI in regulated environments:
- Start with a member problem (e.g., slow call center response, high fraud losses, manual loan reviews)
- Bring compliance and risk into the first conversation
- Map which regs apply: ECOA, FCRA, BSA/AML, privacy, third‑party risk
- Co‑design guardrails before choosing a tool or model
When compliance is a design partner, three things happen:
- You reduce rework and project delays
- You build AI that’s explainable from day one
- You create stronger documentation for audits and examinations
Practical AI use cases that fit inside existing rules
There are plenty of AI use cases that sit comfortably within today’s regulatory framework and still create meaningful value:
- Member service automation: AI chat or voice assistants that handle balance inquiries, card replacement, branch hours, basic FAQs—no credit decisions, no high‑risk activities
- Fraud detection: Machine learning models that flag unusual transaction patterns for human review, not auto‑deny
- Back‑office automation: Classifying documents, extracting data from income statements, routing tickets—reducing manual work without touching eligibility decisions
These “low‑friction” use cases are often the best way to build credibility with your board, regulators, and members that AI can be safe, compliant, and member‑friendly.
Advocacy in Action: What Credit Unions Should Be Asking For
NAFCU’s culture, as Berger describes it, is built around “extreme member service”—to credit unions themselves. That mindset actually translates directly into what credit unions should push for in AI and data policy.
You need clarity more than you need new rules
Most credit unions don’t want lighter regulation; they want clearer regulation. For AI, that means advocating for:
- Clear expectations for explainability: What level of model transparency satisfies fair lending and UDAAP requirements?
- Safe harbors for responsible experimentation: Protection when you test AI in controlled pilots with strong governance
- Reasonable third‑party risk expectations: Guidance that recognizes smaller institutions can’t build everything in‑house
When industry voices are aligned on these points, advocates like NAFCU have a stronger story to tell regulators and lawmakers.
Use member stories to strengthen advocacy positions
Policy makers respond better to stories than technical diagrams. Your real member experiences are advocacy fuel:
- Members waiting 5–7 days for manual underwriting
- Elderly members being targeted by fraud and scams
- Members stuck on hold during peak hours because staffing is thin
Each of these is a member-centric banking problem that AI can help solve—if regulation supports responsible use. Sharing these stories with your league, NAFCU, or other advocacy partners helps shape smarter policy.
Member-Centric AI: Where Compliance, Culture, and Tech Meet
Dan Berger often highlights how credit unions set themselves apart: cooperative structure, community roots, and mission-driven culture. AI should amplify that, not dilute it.
Design AI around member trust, not just efficiency
The fastest way to sabotage your AI strategy is to chase cost savings and forget about trust. For credit unions, trust is the brand.
A member-centric AI program:
- Explains what’s happening in plain language
- Makes it easy to opt out of automated decisions when reasonable
- Uses member data proportionally and transparently
- Avoids “black box” outcomes for high‑impact decisions like lending
If your AI vendor can’t support explainability and clear member communication, that’s a red flag.
Where AI fits best across the member journey
Think of the member journey in four phases and match AI use accordingly:
-
Awareness & onboarding
- Intelligent FAQs and chatbots on your site
- Pre‑qualification tools with clear, compliant disclosures
-
Everyday banking
- Personalized alerts and financial wellness nudges
- Smart routing in contact centers to cut wait times
-
Credit & lending
- Assisted underwriting: AI as a recommendation engine, humans make the final call
- Automated document collection and income verification
-
Protection & retention
- Real‑time fraud and scam detection with human follow‑up
- Churn prediction models that trigger proactive outreach
Notice a pattern: AI handles pattern recognition, predictions, and repetitive work. Humans handle empathy, judgment, and exceptions.
Building an AI Roadmap That Regulators and Members Can Support
NAFCU is focused on creating an environment where credit unions can grow. Your internal roadmap should mirror that focus: sustainable growth, not one‑off AI experiments.
A simple AI roadmap framework for credit unions
Here’s a practical sequence I’ve seen work:
-
Clarify your philosophy
Define your stance on AI in 1–2 paragraphs. How will you use it to support members, not just efficiency? -
Inventory current processes
List processes that are:- Highly manual
- High volume
- Rules‑based or pattern‑based
These are prime candidate areas for AI and automation.
-
Score use cases using four lenses:
- Member impact
- Regulatory complexity
- Data readiness
- Implementation effort
-
Pick 2–3 flagship initiatives
For example: AI chatbot for member service, fraud anomaly detection, and document automation for lending. -
Stand up AI governance
Even a small CU can do this:- Cross‑functional AI committee (IT, compliance, lending, operations)
- Model inventory and risk ratings
- Policies for monitoring, bias checks, and vendor oversight
-
Document everything
When examiners visit, you want to show:- Clear decision logs
- Member communication templates
- Testing results and controls
Where advocacy & education plug into this roadmap
This is where Berger’s comments on NAFCU’s role—representing, assisting, educating, informing—line up beautifully with AI adoption:
- Representing: Your roadmap gives advocates concrete examples of responsible AI use
- Assisting: NAFCU and similar groups can translate new AI guidance into practical checklists
- Educating: Webinars, conferences, and training on AI risk, fair lending, and model governance
- Informing: Regular legal and regulatory updates specific to AI and data use
If you’re not actively feeding your questions and use cases into these channels, you’re leaving value on the table.
Where to Go Next: From Concept to Member Outcomes
The reality? AI for credit unions isn’t about chasing the latest tech trend. It’s about making advocacy, compliance, and member-centric banking work together instead of in separate silos.
Berger’s goal of a regulatory environment where credit unions thrive is achievable, but only if individual institutions:
- Treat compliance as a design partner for AI, not a roadblock
- Give advocacy teams real, member‑driven stories and use cases
- Build AI around trust, explainability, and cooperative values
If you’re serious about AI in 2025, the next step is simple: bring your compliance lead, your advocacy contacts, and your member experience owner into the same room and ask one question:
“Which member problems are we willing to solve with AI—and how do we do it in a way we’d be proud to defend in front of a regulator, a lawmaker, and a member?”
Answer that honestly, and you’ll be ahead of most of the industry.