Regulatory reporting is becoming a major testing ground for artificial intelligence, but financial institutions are struggling to move beyond experimentation.
New research from Regnology, a RegTech provider, suggests that the biggest hurdle is no longer whether firms are willing to test AI, but whether they are prepared to trust it within live reporting processes.
The findings come from responses gathered from 276 practitioners across 22 countries, with most respondents working at financial institutions. Regnology’s 2026 research found that 71% of organisations are either exploring or piloting AI for regulatory reporting. However, only 16% said the technology is already embedded within their operations.
The divide is not simply a question of institutional size. While larger organisations are more likely to have reached the experimentation stage, they are not significantly further ahead when it comes to using AI in production. Among the largest firms surveyed, 49% are piloting AI, compared with only around 8% that have embedded the technology. Across different institution sizes, embedded adoption remains broadly between 8% and 15%.
Regnology describes this difference as the “agentic gap”, referring to the distance between testing an AI application and allowing it to operate as part of the reporting cycle. The research suggests that moving across this gap requires more than investment in technology. Firms need controlled processes that can demonstrate how AI decisions are made, monitored and reviewed while retaining human accountability.
The research indicates that financial institutions are primarily approaching AI as part of wider modernisation programmes rather than treating it as an objective in its own right. Early investment is concentrated on areas such as data analysis and workflow automation, while data quality, modernisation requirements and the potential economic benefits are influencing where firms are applying the technology.
There is also a potentially significant financial incentive for institutions to progress. Regnology estimates that agentic workflows could address approximately 15% to 25% of regulatory reporting expenditure. The opportunity is concentrated in a limited number of high-value activities where significant manual work remains.
However, the research identifies several issues that could prevent firms from scaling these applications. Explainability and auditability remain important considerations, while financial institutions must also translate specialist regulatory knowledge into systems that can operate reliably. Uncertainty around supervisory expectations adds another layer of complexity.
The report also argues that technology alone cannot provide the foundation for wider automation. Strong data quality and governance need to be established before institutions can safely increase the level of authority given to AI within reporting workflows.
Regnology’s research sets out a framework intended to help organisations assess their readiness and move from experimentation towards controlled production. It examines how firms can compare the potential value of individual use cases with their readiness to deploy them, as well as how the level of AI authority should correspond to the nature and risk of each workflow.
The report is aimed at senior professionals responsible for regulatory reporting, data, risk, compliance, finance, technology and operations, alongside transformation leaders and supervisory authorities involved in AI adoption.
The findings point to a regulatory reporting market where experimentation is already widespread, but trusted deployment remains comparatively limited. For financial institutions, the next stage of AI adoption may therefore depend less on increasing the number of pilots and more on establishing the controls needed to put successful applications into production.
Read the Regnology analysis and report
Copyright © 2026 FinTech Global
