AI governance depends on the ability of stakeholders to understand, audit, and oversee AI model decisions. Explainability is the technical and documentation property that enables this oversight. When model cards fail to adequately document explainability, the entire governance chain is compromised.
Why B is Correct: According to ISACA AAIR, inadequate explainability in model documentation is the greatest governance risk because it prevents risk practitioners, auditors, regulators, and business owners from understanding why a model produces its outputs. Without explainability, discriminatory or erroneous decisions cannot be identified, challenged, or corrected. This undermines accountability, compliance, and responsible AI governance at the enterprise level.
Why A is Wrong: Regulatory filing delays represent a compliance timing issue that can be remediated. While risky, they do not fundamentally compromise the governance capability of understanding and overseeing AI behavior.
Why C is Wrong: Decentralized version control creates configuration management challenges and audit trail gaps. These are significant but can be remediated through governance process improvements. Explainability gaps affect the underlying ability to govern the model itself.
Why D is Wrong: Overly detailed technical specifications represent a documentation quality issue that may reduce usability but does not create a governance risk. Excessive detail is easily distilled; absent explainability cannot be reconstructed after the fact.
