Notifications
Clear all
Topic starter
24/06/2026 8:43 pm
TL;DR: Trustworthy AI depends first on governed, high-quality data, according to Collibra, with SAP Business Data Cloud and Collibra positioned as a combined fabric for lineage, semantics, access control, and quality enforcement. The deeper point is that AI trust collapses when the underlying data plane is fragmented, inconsistent, and untraceable.
NHIMG editorial — based on content published by Collibra: The fastest path to trusted AI: Turning high-quality data into high-quality intelligence with SAP Business Data Cloud and Collibra
Questions worth separating out
Q: How should teams govern AI systems that depend on multiple data sources
A: Treat governance as a data-to-decision chain.
Q: Why do data quality problems become security problems in AI programmes
A: Because bad data changes what the system can infer and what it may do next.
Q: How can identity teams support trusted AI without owning the model stack
A: By owning the controls that determine who can access data, how data is classified, and whether provenance can be proven.
Practitioner guidance
- Bind AI use cases to governed data products Require every production AI initiative to reference an approved data product with documented semantics, lineage, and access rules before it can move beyond pilot status.
- Make data lineage part of access governance Include lineage evidence in access reviews for datasets that feed AI models so reviewers can see who touched the source, how it changed, and whether policy remained intact.
- Enforce quality gates before model consumption Block AI pipelines from reading datasets that fail defined quality thresholds, especially where missing values, duplicated records, or inconsistent definitions would alter business decisions.
What's in the full article
Collibra's full blog post covers the operational detail this post intentionally leaves for the source:
- How SAP Business Data Cloud preserves business semantics as data moves across SAP and non-SAP systems.
- How Collibra operationalizes cataloging, lineage, access control, and data quality in one governance layer.
- How the combined approach supports governed data products for AI use cases moving toward production.
- How teams can modernize SAP BW landscapes without losing governance visibility across downstream AI use cases.
👉 Read Collibra's analysis of SAP Business Data Cloud and AI governance →
AI governance and data quality: what IAM teams need to watch?
Explore further
25/06/2026 5:28 am
AI governance still begins with data governance, not model governance. Collibra's argument is directionally right because AI systems inherit trust from the quality, traceability, and policy boundaries of the data they consume. If the data plane is fragmented, explainability becomes retrospective storytelling rather than operational control. Practitioners should treat governed data as the control surface for trustworthy AI.
A few things that frame the scale:
- 67% of organisations still rely heavily on static credentials despite the risks they pose to agentic AI deployments, according to the 2026 Infrastructure Identity Survey.
- Only 44% of organisations have implemented any policies to manage their AI agents, despite 92% agreeing that governing AI agents is critical to enterprise security.
A question worth separating out:
Q: What should organisations measure before moving AI into production
A: They should measure whether the datasets used by AI are complete, consistent, traceable, and governed by enforceable policy. If those signals are weak, production AI will amplify ambiguity instead of reducing it. Governance evidence should be stronger than model confidence.
👉 Read our full editorial: High-quality data is the real control plane for trusted AI
