59% say AI is already built into their core compliance work, but fewer than four in 10 have a named owner for AI risk or a full list of their AI models.
CUBE, a London company that sells AI software for tracking and acting on regulatory changes, commissioned a survey of more than 2,000 compliance and risk managers at banks, insurers and professional services firms. Only 34% said they are confident they could show a regulator how they govern their AI. 59% said AI is already built into their core compliance work, including 10% who let it act autonomously.
Key Takeaways
- 59% say AI is built into their core compliance work.
- 58% expect regulators to scrutinize their AI use more over the next year.
- 57% agree AI’s outputs are not yet trusted enough for critical compliance decisions.
- 34% are confident they could show a regulator how they govern their AI today.
- 10% let AI act autonomously in compliance work.
Most compliance teams rank their AI use as advanced
59% placed their compliance team at one of the survey’s three most advanced AI stages:
- 28% integrated AI into core processes
- 21% said AI is driving strategic risk or compliance decisions
- 10% report that AI is acting autonomously in compliance work
23% use AI for specific tasks. 16% are still testing it.
34% say they could show regulators their governance
34% said they were confident they could show a regulator today how they govern their AI. The figure was 43% in the U.S.
Respondents gave similar answers about specific governance steps. Between 34% and 38% said their firm has each of these AI governance steps in place:
- A named owner for AI risk, such as a senior manager: 38%
- AI results the firm can explain to a regulator and back up with evidence: 38%
- AI risk included in model-risk reviews, with an independent check: 37%
- People who oversee and can challenge decisions made with AI: 36%
- A complete list of the AI models used in risk and compliance: 34%
3% said their AI governance is not yet formal.
57% say AI isn’t trusted for critical decisions
57% agreed that AI’s outputs are not yet trusted enough for critical compliance decisions. Another 25% were neutral.
Most expect more scrutiny and more oversight work
58% expect regulators to scrutinize their use of AI in compliance and risk more over the next 12 months. 44% expect a little more scrutiny and 14% expect a lot more. 28% expect no change, and 14% expect less.
57% expect using AI to add to their governance and oversight work. The figure was 67% in the U.S. and 63% at firms with more than 100,000 employees.
Executives in earlier surveys reported similar gaps
In an October 2026 financial risk survey by Parker &Lawrence Research and RegTech, 69.3% of firms scored higher on AI use than on AI oversight, on the researchers’ 0-to-100 scales.
In a Grant Thornton survey of 950 senior leaders across 10 industries, published in May, 78% doubted their companies could pass an independent AI governance audit within 90 days.
Financial regulators’ global body finalized AI guidance
The Financial Stability Board coordinates financial regulators from 24 countries and jurisdictions. It published 12 final AI practices for financial firms on Oct. 15, three months after CUBE’s survey closed. The FSB calls them a toolkit that firms can tailor to their own risks. It strongly encourages boards and senior managers to refer to them when building AI governance.
Under the practices, boards and senior managers should:
- make sure AI fits the firm’s strategy and risk appetite
- set clear roles and responsibilities for AI
- keep people overseeing AI decisions
Clear responsibility for AI and human oversight are both on the survey’s list of steps that fewer than four in 10 respondents reported. The FSB says it will update the practices as needed.
CUBE surveyed 2,000 compliance and risk managers at financial institutions in June and July 2026. 79% of respondents work in finance and banking, 13% in professional services, and 8% in insurance, in 11 countries, including the U.S., the UK, China, and India.
