According to recent research by Ataccama, 99% of financial institutions are piloting artificial intelligence (AI) but only 3% have succeeded in deploying it at enterprise scale.
The findings come from Ataccama's Financial Services Data Trust Report 2025, which surveyed 300 data leaders from financial organisations in the UK, US and Canada. The report highlights the persistent challenge that while nearly all institutions are testing AI, transitioning these projects into wide-scale production remains rare, largely due to concerns surrounding the quality and reliability of data.
Barriers to AI deployment
Unlike popular assumptions that technological complexity holds back AI expansion, the report identifies data quality and trustworthiness as the principal obstacles. Financial institutions often manage vast networks of legacy systems accrued over decades through mergers, frequently encountering inconsistencies in basic definitions such as "customer", "account" and "transaction". For example, a "closed account" in one system may still be marked as active in another, leading to data mismatches that affect processes including credit decisions and compliance reporting.
While developing an AI model can take only weeks, compiling, cleaning and validating the data required can take months. This mismatch means AI proof-of-concept projects rarely move into full production, pushing Chief Data Officers to reconsider their approaches to data management.
"The low percentage doesn't reflect failure; it reflects reality," said Mike McKee, CEO of Ataccama. "Many financial institutions are racing to deploy AI, but the ones that succeed will find the right balance between speed and trust. That means cleaning up the data, establishing clear ownership, and making sure AI systems can explain their decisions. That balance is what earns both customer trust and regulatory compliance."
The report found that almost half (46%) of surveyed financial services executives ranked improving data quality as their top priority, with over a third identifying it as the biggest challenge. Effective AI at scale, the report asserts, depends less on the sophistication of models and more on ensuring data is consistent, validated and ready for use across business operations.
Regulatory shift
The survey also points to an evolving mindset regarding regulatory compliance. Nearly 50% of financial data leaders now place compliance and reporting high among their business priorities, which is nearly double the rate observed in other industries. New regulations, such as the EU AI Act introducing penalties in 2026 for high-risk AI implementations, are intensifying scrutiny over the accuracy and consistency of data.
Rather than deploying isolated fixes, financial institutions are embedding data governance and quality into daily processes. The focus is moving from cleaning data after the fact to ensuring its reliability and compliance from the outset.
This shift has practical impacts beyond operations. Inaccurate transaction histories can trigger false fraud alerts, while incomplete know-your-customer documentation may delay onboarding or result in regulatory attention. Institutions are integrating quality checks, ownership and observability into the core systems driving their business to address these risks proactively.
McKee added, "Enterprises have evolved from CIO-driven data warehouses and projects to CEO-driven data products and priorities that directly impact competitive advantage and business outcomes. Every decision now depends on knowing which data you can trust, where the gaps are, and how to close them. That means getting the fundamentals right - from KYC documentation to financial records and transaction information - because accuracy isn't optional. A single mismatch can mean a compliance violation or a denied customer request. Perfect data everywhere isn't realistic, but perfect data where it counts is non-negotiable. The institutions that focus on those critical use cases and get them right will scale AI with confidence because they've built trust into every decision their business makes."
Competitive advantage
As AI becomes central to financial strategies, the ability to secure trustworthy, explainable, and well-governed data is emerging as a new competitive differentiator. The report suggests that success with AI requires close coordination between automation, governance, and compliance discipline. Institutions that achieve this can expect improved product delivery times, more accurate regulatory reporting, and stronger confidence in AI-driven operations such as fraud detection, credit assessment, and risk modelling.
Ataccama's research emphasises that there is no AI strategy without a data strategy. Institutions that invest in building robust data foundations are better positioned to meet transparency and accountability standards imposed by new regulations and to harness AI's potential for business outcomes.
