A model risk officer reviews a vendor's documentation for a purchased AI model and finds no information about the training data's demographic composition. What risk management action is MOST appropriate?
Select an answer to reveal the explanation.
Short Explanation
If the vendor’s documentation doesn’t show who was in the training data, you’ve got a real bias risk. Ask the vendor for the demographic composition, and if they can’t provide it, run your own bias testing before you let the model affect real decisions.
Full Explanation
Unknown training-data demographics are a significant model risk gap. The most appropriate action is to request demographic composition data from the vendor and, if unavailable, conduct independent bias testing across demographic subgroups before deployment in decisions affecting individuals. Proceeding because absence of documentation is standard ignores a material risk gap. Requiring the vendor to retrain on a balanced dataset may be a contractual condition, but it is not the immediate action. Limiting use to documented groups is operationally unworkable because the composition is unknown.