One of the most revealing findings in York Public Relations’ 2026 State of AI survey challenges a common assumption about what is holding banks and credit unions back.
Trust in AI outputs and accuracy was selected as a top adoption barrier by 44% of respondents. Regulatory and compliance risk was selected by 32%.
In other words, respondents are more concerned about whether AI can be trusted than whether it can be regulated.
From “Can We Use It?” To “Can We Trust It?”
That distinction signals an important evolution in the AI conversation.
Financial institutions increasingly understand that AI can be used. The harder question is whether the output is consistently accurate, explainable and reliable enough to use at scale.
That becomes particularly important in high-stakes areas such as lending, fraud detection and compliance, where an incorrect output can carry financial, regulatory and reputational consequences.
Trust Is A Product Requirement
For AI providers, trust cannot be treated purely as a communications issue.
Institutions need evidence.
That can include demonstrated accuracy, transparent governance, explainability, audit trails, clearly defined human oversight and controls around how institutional data is used.
The goal is not to convince buyers to ignore AI’s limitations. It is to show them how those limitations are identified, managed and governed.
Data Matters, Too
Thirty percent of respondents cited data quality and integration as a barrier.
That figure may understate the challenge. Institutions often discover fragmented, siloed or inconsistent data only after implementation begins.
Poor data makes reliable AI harder. And unreliable AI makes trust harder.
The path to greater AI adoption therefore depends on more than better models.
It depends on creating systems institutions can understand, verify and confidently govern.