Whole Of Case Visibility
Whole of Case Visibility is a governance principle mandating comprehensive transparency and forensic-level auditability for artificial intelligence systems used in high-stakes decision-making. This framework requires that every stage of an algorithmic workflow, from initial data input to final output, be fully documented and traceable. The standard ensures that decisions influenced by automated systems can be rigorously examined, preventing opaque processes from obscuring critical errors or biases.
The concept emerged as a direct response to the wrongful arrest of Faarax Jama, a Somali-American man detained for over 12 hours due to a facial recognition error. The incident highlighted the dangers of relying on black-box algorithms in law enforcement without adequate oversight or the ability to reconstruct the specific data points and logic that led to a false match.
Implementing this visibility requires organizations to maintain detailed logs of data provenance, model versioning, and decision thresholds. By establishing a clear chain of custody for algorithmic decisions, stakeholders can identify where failures occurred, whether in data quality, model training, or deployment parameters, thereby enhancing accountability and reducing the risk of systemic harm.
Source Notes
- 2026-04-14: I Looked At Amazon After They Fired 16,000 Engineers. Their AI Broke Everything.