AI trust gap threatens banks’ responsible adoption

Financial institutions confront rising risks from untested and opaque AI systems.

Financial institutions across Asia are facing a growing “trust dilemma” in their AI strategies, as confidence in AI outputs increasingly diverges from the technology’s actual trustworthiness. According to Patrick Xhonneux, Senior Vice President, Marketing at SAS, the widening gap is putting banks at risk of regulatory breaches, reputational damage, and stalled innovation.

In an interview conducted at the sidelines of Singapore Fintech Festival 2025, Xhonneux said new research from IDC and SAS shows the problem is far-reaching. “We see that 47% of companies in AP are facing that trust dilemma,” he said. The issue stems from a mismatch between how much companies rely on their AI systems and how reliable those systems truly are.

Xhonneux emphasised that trust must now be treated as a measurable performance indicator. “Trust is the new currency of AI value creation,” he said, pointing to fairness, transparency, and bias detection as essential components. SAS has introduced model cards—“a bit like the food nutrition labels”—that display whether a model “is fair… transparent [in] how it has been trained, validated and deployed.”

Regulatory scrutiny is intensifying across Asia, further elevating the stakes. “MAS is really making sure that banks can explain… the model deriving a specific outcome,” he noted. Strong governance frameworks will be critical to developing “sustainable and long term AI models deriving outcomes.”

When advising institutions on responsible scaling, Xhonneux stressed the importance of knowing their AI maturity level. SAS uses an AI governance assessment model that classifies organisations as reactive, co-active, or leaders. “If you know where you are and you start with a governance framework that integrates data, quality, transparency… then you're sure that you will be able to scale up your AI model and get a higher return on investment.”