Topic starter
24/06/2026 7:08 pm
TL;DR: Data quality is now the biggest barrier to successful AI projects for 44% of organisations, according to BARC, and Collibra’s 2026 scorecard frames trustworthy data as the precondition for production AI, GenAI features, and governed decision-making. The real issue is not cataloging alone, but whether governance can enforce trust where data is actually consumed.
Editorial analysis by NHI Mgmt Group, based on content published by Collibra: “Three years a Market Leader. In 2026, Collibra came out on top”.
Key questions
Q: How should teams govern AI assets and data together without creating duplicate controls?
A: Build a single governance inventory that links datasets, models, policies, and approvals, then reuse governance artefacts only when ownership, versioning, and review evidence remain intact.
Q: What is the risk of treating AI governance as a documentation exercise?
A: Documentation alone does not stop an AI system from using stale, unclassified, or improperly accessed data.
Q: How do organisations know whether AI governance is actually working?
A: AI governance is working when teams can prove that data access, identity permissions, and runtime controls line up with policy in practice.
Practitioner guidance
- Enforce policy at the point of data use Move beyond cataloguing and ensure access rules are enforced inside the cloud data platforms where AI retrieves information, so policy decisions affect live consumption rather than later review.
- Extend governance to AI assets and datasets Treat models, agents, and training datasets as governed enterprise objects with ownership, classification, and lifecycle controls, not as informal artefacts outside the data governance programme.
- Map unstructured data into governed workflows Bring raw files, documents, recordings, and emails into the same governance model as structured data so AI cannot bypass controls by switching source type.
Bottom line: AI governance depends on trustworthy data at the point of consumption, not just on catalogues and stewardship records.
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
25/06/2026 4:17 am
Trustworthy data has become the enforcement surface for enterprise AI governance: the old assumption was that governance could sit around the edges of analytics and documentation. That assumption fails when AI consumes data directly at runtime and turns context into action. The implication is that governance programmes must be judged by enforcement at the point of use, not by catalogue completeness alone.
A question worth separating out:
Q: Should data teams govern models, agents and datasets together?
A: Yes. Once models, agents, and training datasets operate as enterprise assets, separating their governance from the underlying data creates gaps in accountability and lifecycle control. A unified programme makes it easier to assign ownership, track lineage, and maintain trust across the full path from data source to decision outcome.
👉 Read our full editorial: Data trust is now the control plane for enterprise AI governance