Search the Atlas

Search risks, controls, and glossary terms

MediumDoc GapsRealized

Inadequate model documentation / model cards

Transparency & Explainability

Description

Missing model cards, intended-use statements, and limitations documentation undermine governance and accountability.

Example scenario

An AI system is in production with no documented intended use or limitations for the audit.

Real-world evidenceRealized

Multiple media outlets and commercial services were caught publishing AI-generated content without disclosure in 2022–2023, and regulators including the FTC and EU authorities have since cited specific cases. This crosses from demonstrated to realized because undisclosed AI content reached mass readership before correction.

Primary mitigations

  • Mandatory model cards
  • intended-use & limitation statements
  • documentation gates in MLOps
  • version records.

Detection signals

Documentation-completeness audits; model-card coverage.

Mitigating controls

5
Non-agentic controls

Related risks in Transparency & Explainability