CUBE Content Governance Global Ltd

11/16/2026 | Press release | Distributed by Public on 11/16/2026 03:41

AI governance: from progress to proof

Compliance and risk teams have spent the past year getting faster. The next step is showing the work holds up, and AI governance is where that proof counts most.

CUBE's Cost of Compliance Report 2026 captures the shift. 61% of firms now embed regulatory change in under a year, up from 26% in 2025. Yet only 45% rate their approach as effective, down from 59%. AI follows the same pattern. 59% say it is embedded or advanced in core processes, but only 34% are confident they could demonstrate their AI governance to a regulator today.

We call the distance between what a firm has deployed and what it can prove the evidence gap, and firms are working to close it. In a recent webinar, Olga Cueves Gomez, Managing Director at EY, and Todd Ehret, Senior Regulatory Intelligence Expert at CUBE, joined Cheryl Feldesman, Regional Executive Head, US East North at CUBE to discuss how to close it.

Visibility turns capability into confidence

59% of firms say AI is embedded or advanced in their core processes, and 34% are confident they could demonstrate their AI governance to a regulator today. That 25-point difference is an opportunity. Todd's reading is that firms are probably doing more than they give themselves credit for, and that the low number reflects uncertainty about where AI is actually in use. Large institutions already have rigorous model risk processes, and AI is being folded into them.

Compliance and risk has been here before. Cybersecurity and data privacy arrived as new regimes that required new policies, controls and ways of working. The playbook exists. What AI adds is pace, and a regulatory picture that is still taking shape: the EU AI Act is in effect, and state-level rules in the US are building into a patchwork.

The practical starting point is unglamorous. Build a complete inventory of the AI tools in use and where they sit. Confidence follows visibility.

Why AI governance needs an "expert in the loop", not just a "human in the loop"

This was the idea that stuck with us from CUBE Connect last week. A human who clicks approve adds little but an expert who understands what the AI was asked, what data it drew on and how to challenge the answer adds a great deal.

That brings skills into focus. 57% of respondents don't yet trust AI outputs for critical decisions, and 32% cite talent gaps as a barrier to scaling AI. Training is part of the governance work, and Olga is seeing clients limit which AI platforms staff can use while they build it. A slower, more controlled start often leads to faster progress later.

How does regulatory change insight reach the control?

Olga's observation about regulatory change management is the sharpest in the report's findings. Most of the effort goes into the front end: horizon scanning, alerts and triage. Firms are monitoring more than ever, with 17% now tracking over 100 developments a month, up from 2% a year ago. But the insight often stops before it reaches the inventory, the control or the test. An impact assessment that never answers "is my control effective?" has done half its job.

This is where AI earns its place. It can pull the required action out of hundreds of pages of regulatory text, and it can feed that insight into testing, monitoring and risk assessment through a joined-up data model. As Todd put it, speed without accuracy is a burnt dinner served quickly.

How should firms handle regulatory divergence?

39% of respondents rank deregulatory rollback and divergence among their greatest cross-border risks. Simplification in one jurisdiction doesn't remove obligations in another. During the webinar, both Olga and Todd pointed to a few anchors: the EU AI Act is a useful reference even for non-EU firms, the Financial Stability Board's proposed sound practices for AI adoption offer a benchmark, and a clear view of your cross-border footprint and sanctions exposure matters more as financial crime pressure grows.

Six AI governance actions for the next 12 months

  1. Inventory every AI tool in use, and who owns it. Firms often do more than they give themselves credit for, but they cannot show it until they know where AI sits.

  2. Put experts, not just people, into AI-driven decisions, and train them. An expert understands what the AI was asked, what data it drew on and how to challenge the answer. With 57% of respondents not yet trusting AI outputs for critical decisions, training is part of the governance work.

  3. Set clear boundaries on which AI platforms your teams can use. Olga is seeing clients limit platform access while they build skills. A slower, more controlled start often leads to faster progress later.

  4. Connect horizon scanning to your regulatory inventory and controls. Insight that stops at triage has done half its job. Every impact assessment should answer whether the control is effective.

  5. Treat data governance as shared work between a central data office and the business units. A joined-up data model lets regulatory insight feed testing, monitoring and risk assessment. Without shared ownership, that insight stalls before it reaches the control.

  6. Map your cross-border exposure, and benchmark against the EU AI Act and the FSB's sound practices. Simplification in one jurisdiction does not remove obligations in another. Include sanctions exposure, which matters more as financial crime pressure grows.

Firms that close the distance between what they have deployed and what they can evidence will get the most from AI over the next 12 months.

Frequently asked questions about AI governance

What is AI governance?

AI governance is a firm's ability to show regulators where AI is used, how it is controlled and what evidence supports those controls. In CUBE's Cost of Compliance Report 2026, only 34% of firms were confident they could demonstrate their AI governance to a regulator today.

What is the first step to improving AI governance?

Build a complete inventory of the AI tools in use, where they sit and who owns them. Confidence follows visibility.

How should firms approach divergent AI regulation?

Simplification in one jurisdiction does not remove obligations in another. The speakers pointed to the EU AI Act as a useful reference, the Financial Stability Board's proposed sound practices as a benchmark, and a clear view of your cross-border footprint.

Download the Cost of Compliance Report 2026 to see the full findings, or let's talk about what this means for your team.

Source: CUBE, The Cost of Compliance Report 2026.

CUBE Content Governance Global Ltd published this content on November 16, 2026, and is solely responsible for the information contained herein. Distributed via Public Technologies (PUBT), unedited and unaltered, on November 16, 2026 at 09:42 UTC. If you believe the information included in the content is inaccurate or outdated and requires editing or removal, please contact us at [email protected]