The relative novelty of artificial intelligence use within financial services has been highlighted in a global study of the technology’s adoption, with many data challenges remaining and blind spots in security still widespread.
While the 2026 Global AI in Financial Services Report by the Cambridge Centre for Alternative Finance found firms were eager to integrate AI into their operations, it also underlined how a large proportion is either finding it difficult to achieve their aims or isn’t putting the best safeguards in place.
Data quality, talent shortages and legacy systems continue to be sticking points in the effective rollout of AI programmes, and its safe development in the sector may be hindered by a weak regulatory regime, the report’s authors concluded.
As well, the gap between those viewing the technology as ultimately beneficial and those who see it as a source of risks remains narrow; few reported substantial productivity gains after implementing AI tools but many fear a greater level of fraud is possible with them in place.
“This transition promises to enhance operational efficiency, unlock new value and reshape market dynamics,” the authors of the report wrote. “Yet, as in earlier phases of financial innovation, this rapid pace of technological change risks outpacing organisational readiness and regulatory capacity.”
Gung Ho for AI
The survey of 628 respondents in 151 countries across banking, insurance, payments, capital markets, digital assets and other financial services sectors found no end of enthusiasm for AI’s use. Four-fifths of respondents said they are adopting the technology at some level, and two-fifths said they are at an advanced stage of rollout. Their focus has largely been on generative AI and machine learning, while four-fifths said they were looking to adopt autonomous agentic AI in the future.
Use cases are concentrated in the back office, with process automation, data visualisation and data knowledge management topping the list. Just one-fifth said they were using it to rebuild their business model.
The most benefits perceived from AI use were in technology and data functions, where four-fifths said they saw improvements. But fewer respondents said they were seeing a correlated boost to profitability; two-fifths said they had, while an equal proportion said they saw no change. Higher spending on AI appears to bring the greatest impact, the survey found – two-thirds of those who invested more than US$100,000 on development said they saw a profit boost, while that figure fell to two-fifths among those who had spent less.
Same Old Problems
Perhaps unsurprisingly, the authors said that the greatest challenges to successful AI rollout were those that have afflicted financial institutions for many years.
Poor data quality and availability were a hindrance to two-thirds of AI technology vendors questioned, half of regulators and two-fifths of institutions. Vendors also found “acute data-related challenges” among their clients, with almost three-quarters reporting shortcomings in data quality and completeness and a half citing legacy systems and data silos.
Structural challenges also lay in the lack of preparedness of regulators to oversee AI’s development. Only 20 per cent of them said they were in advanced stages of AI development themselves. Nevertheless, they are optimistic about AI’s role in achieving their oversight goals over the next five years.
Off the Guardrails
Poor safeguards among companies could further derail deployment. The survey found surprisingly few respondents had built explainability and bias corrections into their models. This could be connected to the fact that more saw risks of fraud, deepfakes and other menaces from the adoption of AI.
Regulators are not the only participants lacking preparedness. The report points to a growing challenge posed by a shortage of talent to guide AI rollouts and operate alongside the models. Only one in 10 respondents said their workers were highly prepared.
The authors said that the previous report, published in 2020, highlighted this as a critical barrier to AI development and noted that the situation hadn’t improved.
“Critically, workforce preparedness is a leading indicator for positive financial returns from AI,” the authors wrote. “Institutions that reported positive AI profitability outcomes were nearly four times more likely to describe their workforce as highly prepared (23 per cent) compared with firms reporting no change in profitability (6 per cent).
“This strongly indicates that extracting value from AI depends on both human capital investment as well as technology procurement and adoption.”
Solutions at Hand
In many ways, the report’s findings were unsurprising – many of today’s shortcomings existed five years ago. Wayne Kiphart, chief executive at CloudFirst Global, which helps organisations overcome the limitations of legacy systems, observed that there are steps that organisations can take to mitigate them.
“As AI is increasingly embedded into financial services, the real risk is not the presence of legacy systems, but the lack of expertise and oversight needed to integrate and manage these technologies securely,” Kiphart told Data Management Insight. “Without the right governance and skills in place, organisations risk introducing new vulnerabilities into environments that are otherwise proven, resilient, and business-critical.”
Subscribe to our newsletter
