Which statement best summarises when a governed AI model should be reassessed?
Select an answer to reveal the explanation.
Short Explanation
You reassess when data, purpose, version, or risk conditions materially change, not just when a release calendar says so. Don't wait for a major version or convenience; material change is the trigger.
Full Explanation
The correct statement is the one that says reassessment should happen whenever the model's data, purpose, version, or risk conditions materially change, rather than only at fixed intervals or never after initial approval. Tying reassessment to material change keeps governance responsive to what actually affects the model's behaviour and risk. The statement that reassessment should occur only when a new major version is released is incorrect because minor changes can still be material, particularly to data or risk conditions. The statement that reassessment should occur only once, at initial deployment, is incorrect because treating deployment as the endpoint of governance leaves the model unmonitored. The statement that reassessment should occur whenever convenient for the development team's release schedule is incorrect because release convenience is unrelated to whether a material change has occurred. For the exam, state the trigger as material change across data, purpose, version, or risk, as a complete set.