A court AI audit plan lists fieldwork tasks but maps no criteria to AI policies, standards, or regulations. What should the auditor correct?
Select an answer to reveal the explanation.
Short Explanation
Fieldwork without criteria is a scavenger hunt, not an audit. You need mapped policies, standards, or rules—not court case outcomes or judicial efficiency targets. A vendor's own checklist or general IT control frameworks for court systems operations isn't named AI criteria either.
Full Explanation
Selecting and mapping audit criteria to applicable AI policies, standards, or regulations works because fieldwork without criteria cannot support objective conclusions. Court AI controls and outcomes must be evaluated against named expectations—judicial AI use policies, data protection rules, model documentation standards, or adopted frameworks. Mapping criteria to tasks turns a scavenger hunt into an audit and clarifies what “pass” means for sensitive justice-adjacent systems.
Mapping audit criteria to court case outcomes and judicial efficiency targets fails: outcomes and efficiency targets are results or goals, not the named AI policies, standards, or regulations needed to judge court AI controls.
Relying on the vendor's own testing checklist as the audit criteria lets the vendor grade its own homework instead of holding the court AI to independently adopted policies, standards, or regulations. Assessing compliance using general IT control frameworks for court systems operations still leaves fieldwork without criteria specific to court AI's justice-process and regulatory risks, so a 'pass' against a general IT framework does not mean the system meets what the court actually requires.
Exam caveat: criteria must be appropriate and complete enough for the engagement objective—over-narrow criteria can greenwash a risky court AI that meets a weak checklist. Operational check: in planning, list criteria sources (internal AI policy, statutes, adopted standards), map each fieldwork procedure to criteria IDs, and retain that matrix in the court AI audit file before testing begins.