Good Morning!
Last week, we looked at how fraudulent documents are already inside your loan pipeline. This week, it goes one layer deeper.
The fraud hitting credit unions right now is bad identities built to pass your checks completely. New data out this week puts a number on how bad it has gotten.
And the industry's biggest technology conference wrapped in Dallas with one clear message for every CU still sitting on the sidelines.
Let’s get into it.
TRENDING AI NEWS FOR CU
Credit union leaders at EDGE26 this week: AI adoption is not the question anymore
At the EDGE26 Technology Conference in Dallas this week, Trellance CEO Tom Davis told a room of credit union executives that the industry has moved past asking whether to adopt AI. The conference drew CU executives, fintech leaders, and CUSO partners from across the country to discuss how institutions are operationalizing AI across lending, fraud, and member experience
Why it matters for your CU: The conversation at the industry's leading technology conference has shifted from "should we" to "how fast and how governed." If your leadership team is still in evaluation mode, your peers at the table this week are already in production.
New NCUA board nominee, new exam priorities
John Crews has been nominated to the NCUA board, arriving midstream amid 11 deregulation rounds, with AI governance, vendor oversight, and model risk guidance all being actively rewritten. Board composition shapes what examiners prioritize.
Why it matters for your CU: Board nominees do not just shape policy. They shape what questions show up in your next examination. This one arrives at the most consequential moment for AI oversight in NCUA history.
Fraud rates climbed at 67% of financial institutions in 2025. Synthetic identity is the reason.
Equifax research published this week shows that synthetic identities, fake profiles built from real and fabricated data, are now bypassing traditional credit checks at scale. A standard consumer laptop can generate fraud-ready fake identities in minutes. The barrier to entry for organized fraud has effectively collapsed.
Why it matters for your CU: Synthetic fraud does not show up as fraud on your reports. It shows up as a charge-off six to eighteen months later. By then, the money is gone, and the identity never existed.
DEEP DIVE
Your staff is using AI you have not approved. Here is what that costs you.
Think about what your team processes daily. Loan applications. Tax documents. CECL inputs. BSA/AML case files. Member PII across every origination touchpoint.
Do you know how AI is touching any of it, including tools staff found on their own?
AI confidence among community financial institution leaders jumped from 17% to 50% in a single year, according to The Financial Brand. Governance has not kept pace. Most CUs have not mapped which AI capabilities are embedded in their core, their LOS, or their digital banking platform, let alone what staff are accessing outside those systems.
Your loan officers have stipulations to clear. Your ops team has queues. A free AI tool available in 30 seconds moves faster than most approved systems. That gap is where unmanaged AI use grows quietly.
The NCUA's AI Resource Hub ties oversight directly to your existing vendor management framework under Letters 07-CU-13 and 01-CU-20. Your vendors' embedded AI features are your examination responsibility.
The institutions doing this well are not restricting AI. They are channeling it. Governed alternatives reduce unsanctioned tool use sharply. That is the difference between an AI policy and an AI program.
Three things to do this week:
Ask IT what AI tools are on your network, not what is approved, but what is actually in use.
Search your third-party vendor agreements for AI. Confirm whether embedded features are covered under your existing oversight framework.
Pull your acceptable use policy and search for the word AI. If it is not there, you are exposed at your next examination.
FROM MULTIMODAL
Multimodal at EDGE26
Nicholas Bianchi at EDGE26
Nicholas Bianchi demoed AgentFlow live in Dallas to a room full of credit union and community bank leaders. Pre-built agentic AI playbooks for lending, compliance, and servicing, all running directly on the cores your team already uses.
If you missed it and want to see what we showed:
Data point this week
$23B
Projected annual losses from synthetic identity fraud by 2030 are already hitting 67% of financial institutions today.
Source: Deloitte via Equifax
