Good Morning!
June 30 is 19 days away. That's when America's first AI lending law takes effect, and it applies to any bank serving Colorado customers, not just Colorado-chartered institutions.
This week we also got fresh data on how unprepared most banks are when an AI model goes wrong. Both are worth your attention before the weekend.
TRENDING AI NEWS FOR CB
72% of banks are least prepared for the two things regulators will ask for first
Wolters Kluwer's H1 2026 Banking AI Risk and Governance Index, published this week, surveyed 230 banking professionals across community, midsize, and large institutions. When asked which area of AI risk their bank was least prepared for, 72% chose model kill-switch protocols (34%) or regulatory reporting of AI failures (38%). The report stated: "Regulatory reporting and kill-switch protocols are not esoteric capabilities; they are the minimum viable requirements for managing an AI incident in a regulated environment."
Why it matters for your bank: When an AI model produces a bad outcome, regulators will ask two questions first: did you know, and did you stop it? Most banks cannot answer both confidently right now.
Research shows community banks can outrank big banks in AI-generated recommendations
New research from EMARKETER's AI Visibility Index, covered by The Financial Brand this week, analyzed thousands of ChatGPT responses across financial services categories including banking accounts and credit products. The finding: the brands that consistently surface in AI-generated recommendations are not always the largest institutions. The key differentiator is clarity. Banks performing well articulate exactly who they serve, what they offer, and when they should be considered. Complexity and jargon push institutions out of AI answers entirely.
Why it matters for your bank: Community banks have a structural advantage here. A focused geography, a specific customer type, and a clear service area are exactly what AI rewards. The banks that communicate that clearly will show up. The ones that don't will be invisible regardless of asset size.
Visa and OpenAI just put AI agents on payment rails
Visa announced a partnership with OpenAI this week, embedding Visa's payment network directly into OpenAI's platform, enabling AI agents to initiate transactions on behalf of customers within defined spending limits and merchant controls. Tokenized credentials and real-time fraud monitoring underpin the system. It is part of Visa's broader Intelligent Commerce initiative alongside Microsoft, Samsung, and Stripe.
Why it matters for your bank: Agent-initiated transactions will hit your payment authorization systems whether you are ready or not. The fraud models and authorization logic your bank built around human-initiated payments will need to account for AI agents acting on customers' behalf. That planning starts now.
DEEP DIVE
America's first AI lending law takes effect in 19 days. Your bank may already be in scope.
On June 30, 2026, Colorado's Consumer Protections for Artificial Intelligence Act becomes enforceable. It is the first comprehensive state AI law in the United States. The detail most community banks have missed: it applies to any institution serving Colorado customers, not only Colorado-chartered banks.
If your bank has customers in Colorado through digital channels, loan origination, or any online product, and you use AI to influence those credit decisions, you are in scope.
The law covers high-risk AI systems, any system that substantially influences a consequential decision. In banking, that means loan approvals, credit scoring, underwriting, interest rate determination, and access to financial products. The compliance requirements are specific. A documented risk management program. Impact assessments on AI systems that influence credit decisions. Consumer disclosures before an AI-influenced decision is made. Bias testing with documented evidence that you evaluated less discriminatory alternatives. Human oversight on consequential decisions.
Penalties run $20,000 per consumer per violation. If a senior citizen is affected, that rises to $50,000. Enforcement sits with the Colorado Attorney General, who has signaled the office will prioritize consumer complaints over proactive audits. The first cases will likely come from a denied loan applicant who discovers AI was involved.
There is a federal compliance safe harbor. The law states that a financial institution subject to and compliant with federal laws providing substantially similar protections may qualify. But that compliance must be documented. Being subject to federal law is not the same as being in compliance with it.
The OCC, Federal Reserve, and FDIC have separately signaled that an RFI on model risk management, specifically covering generative and agentic AI, is coming soon. Colorado is the opening move. Multiple states are watching.
Three things to do this week:
Check whether your bank originates loans or serves customers in Colorado through any channel. If yes, identify which AI systems touch those decisions.
Pull your model inventory and flag any system influencing credit approvals, pricing, or underwriting for Colorado customers. That is your scope list.
Ask your compliance team whether your current documentation satisfies the Colorado Act's impact assessment and bias testing requirements. If they haven't seen the law yet, send it today.
FROM MULTIMODAL
Purpose-built for community bank lending
We build AI agents and pre-built workflows for community bank lending operations: document intake, classification, data extraction, and decisioning support, all running on the cores your team already uses.
With Claude Fable 5 out this week, many lending teams are asking where a general-purpose AI model fits relative to a purpose-built agent. We wrote up the answer.
➡️ Read here
If you want to see how it works at your bank, a 30-minute demo is the fastest way to find out.
Data point this week
$80.5 billion
Net income across FDIC-insured institutions in Q1 2026, up 14.3% year-over-year and the highest quarterly earnings on record. Community banks are in a strong financial position heading into the second half of 2026.
Source: FDIC Quarterly Banking Profile, Q1 2026
