Large financial institutions are moving faster than smaller peers when it comes to testing AI for regulatory reporting, but size has yet to give them a meaningful advantage in bringing those systems into live use.
Research from Regnology suggests the industry’s biggest challenge is not experimentation, but establishing enough confidence to put AI into production.
Among the largest organisations surveyed, almost half, or 49%, are currently piloting AI. Yet only around 8% have embedded the technology into their operations. That compares with embedded adoption of between 8% and 15% across the other institution size groups, suggesting that greater resources have helped firms test more use cases without necessarily making them more comfortable deploying AI at scale.
Regnology reached its conclusions after gathering responses from 276 practitioners across 22 countries, with most respondents working for financial institutions. Overall, 71% of organisations said they were either exploring or piloting AI within regulatory reporting, while only 16% reported that the technology had become embedded in their operations.
The RegTech provider refers to the divide between experimentation and operational use as the “agentic gap”. Rather than measuring the difference between large and small institutions, the term describes the challenge of moving from an AI proof of concept to a system that can be trusted during an actual reporting cycle.
That transition requires firms to establish safeguards around how AI operates, according to the research. Monitoring, review processes and clear accountability all become increasingly important as organisations give AI a greater role in regulatory workflows.
There is also a sizeable commercial incentive for firms to overcome those hurdles. Regnology estimates that agentic workflows could potentially address between 15% and 25% of regulatory reporting expenditure. The opportunity is concentrated in a relatively small number of activities where manual processes continue to consume significant resources.
Current investment patterns suggest institutions are approaching AI primarily as part of wider efforts to modernise reporting. Data analysis and workflow automation are attracting early attention, while data quality requirements, modernisation programmes and potential efficiency gains are helping determine where AI is being deployed.
However, moving from experimentation to production presents several challenges. Firms must address questions around whether AI outputs can be explained and audited, while also finding effective ways to translate specialist regulatory knowledge into automated systems. Uncertainty around what supervisors will expect from AI-enabled reporting adds another consideration.
The research also places emphasis on the infrastructure supporting AI adoption. Improvements in data quality and governance are presented as prerequisites for institutions that want to increase the level of autonomy given to AI without weakening oversight.
Regnology’s report proposes a framework for assessing potential AI use cases according to both their value and an organisation’s readiness to implement them. It also considers how much decision-making authority should be assigned to AI based on the characteristics and risk profile of individual workflows.
The research is intended for leaders across regulatory reporting, data, risk, compliance, finance, technology and operations, alongside transformation teams and supervisory bodies responsible for AI adoption.
Overall, the findings suggest that AI experimentation in regulatory reporting is becoming widespread, but deployment remains constrained by trust and governance. For financial institutions, the challenge now is to identify which pilots can safely make the transition into production and establish the controls needed to support them.
Read the Regnology analysis and report
Copyright © 2026 FinTech Global
Copyright © 2026 RegTech Analyst
