Study: Only 11% of banks have cracked the code on trustworthy AI

Even as AI spending surges, few banks have established the necessary governance and guardrails – and nearly half misjudge their own AI readiness

New research reveals that while banks are ramping up investment in AI at an unprecedented pace, most are failing to implement the governance and infrastructure needed to earn trust. The majority of financial institutions are still struggling to use AI responsibly and reliably, leaving a gap between ambition and actionable, trustworthy AI practices.

The Data and AI Impact Report: The Trust Imperative, commissioned by AI and data leader SAS, with research insights by IDC, found that almost one quarter (23%) of banks globally operate at the highest level of IDC’s Trustworthy AI Index.

Within this group, only 11% of banks globally have achieved both high internal confidence in AI and AI systems that are demonstrably trustworthy - effectively cracking the code on trustworthy AI. Nearly half (47%) of respondents globally fall into what IDC calls the “trust dilemma” – either under-using reliable AI because they do not sufficiently trust it or over-relying on AI systems that haven’t been adequately validated.

In Europe specifically, banking organisations are moderately advanced in both AI maturity and data infrastructure, but overall adoption remains modest due to strict regulatory requirements that limit how banks can store, use, and move data and the AI use cases they can pursue.

Across the region 21% of banks are allocating resources at a high level to deliver trustworthy AI, while around a third fall into the “danger zone,” reporting high confidence in AI but without matching investments to ensure it is trustworthy.

Commenting on the findings, Dr Iain Brown, Global Head of AI & Data Science at SAS, said:

“Banking leads every sector on trustworthy AI in this study – yet most banks are still some way from operationalising it at scale. Roughly nine in 10 have yet to fully align trust with demonstrable governance, and around one in five are still operating with siloed data foundations. In Europe’s regulatory environment, closing the gap between AI ambition and AI readiness must now become a board level priority.”

Investment is rising, but foundations remain fragile

The report, based on a global, cross-industry survey of 2,375 IT and business leaders, reveals a troubling pattern: investment in AI capabilities is not being matched by investment in the responsible innovation pillars that make AI dependable.

In an industry where a single model failure can trigger regulatory penalties from financial bodies such as the FCA or erode consumer confidence overnight, that’s a dangerous disconnect.

Most banks globally (60%) expect AI spending to grow between 4% and 20%, with 12% anticipating even larger increases. In Europe, the pace is more measured, with most banks projecting at least 4% growth over the coming year. Despite this momentum, foundational weaknesses remain widespread.

Nearly one in five banks worldwide (19%) still operate with siloed data infrastructures, 45% lack effective data governance, 41% do not have a centralised or optimised data platform, and 42% report shortages of specialised AI skills.

To address these challenges, more than half of banks (52%) plan to expand their AI architecture, while 43% are building or growing dedicated AI teams. Fewer than one-third (31%) plan to focus on developing and tuning AI models themselves, underscoring that structural readiness, not ambition, remains the key barrier to trustworthy AI adoption.

"The banking sector clearly understands AI's potential, but understanding and execution are not the same," said Kathy Lange, Research Director of the AI and Automation Practice at IDC. "Without strong data architectures, governance frameworks and talent pipelines, banks risk pouring money into AI initiatives that can't deliver ROI – or worse, that undermine the very trust they depend on."

Responsible innovation, not cost savings, drives AI ROI

The report also challenges the assumption that AI's primary value in banking is cost-cutting. On the contrary, European banks prioritise product and service innovation over process efficiency as the leading source of AI-driven value.

Cross-industry ROI figures show banks are onto something. Organisations using AI to improve customer experience reported the highest return – $1.83 (approximately £1.37) for every dollar invested – followed closely by those centred on expanding market share ($1.74 or approximately £1.30). Those focused on cost savings reported the lowest – $1.54 (approximately £1.15) per dollar.

Moreover, organisations that prioritised trustworthy AI were 60% more likely to report doubling overall return on their AI initiatives. That’s solid proof that responsible innovation is a growth accelerator that more than pays for itself.

Banks are also moving more decisively than other sectors towards agentic AI, with nearly one-third planning increases in trustworthy AI investment to support more autonomous systems. But as AI systems gain greater decision-making authority, the consequences of weak governance grow more significant.

"Regulators are watching. Customers are watching. And right now, nearly half of banks are using unproven AI – or hesitating to tap AI they’ve validated," said Alex Kwiatkowski, Director of Global Financial Services at SAS. “No bank wants to become an ‘also-ran’ in this highly competitive race, and cost-savings alone won’t keep them in it.

“The banks that win will be ones that invest in governance, explainability, transparency and strong data foundations before they scale, not after something breaks.”

Curious how to close the gap between trust and governance? Tune in for the webinar, Banking on AI: The 2026 Maturity Pivot.

Download and explore the full report: 
SAS.com/ai-impact.