Ask an AI vendor where customer data goes, whether it uses that data to train models, and what happens to the data when the agreement ends. For decision rollback, ask who can correct or reverse an AI-driven decision and what steps that process requires. Procurement should review the answers with security and the business owner before approving the purchase.

At ProcureCon IT Sourcing 2026, Amy Babcock, Senior Manager – Technology Strategic Sourcing at Nordstrom, encouraged procurement to work with engineering on contingency plans when AI capabilities change. She asked: "Do they have a backup plan? Can they easily port this over to another solution?" Her comments give buyers another question to resolve with their technical teams: how they would continue operating if a supplier withdrew a capability they depended on.

In another 2026 discussion, a procurement practitioner described requesting AI disclosures that covered data destinations and decision rollback. She used an insurance example to explain why buyers should understand how to reverse decisions when regulatory requirements change.

Request written answers tied to your intended use case, then ask the supplier to demonstrate how an authorized user would correct a decision. Have security review data handling and model training permissions, and work with legal to document the agreed responsibilities. Include engineering in the review of backup options so the business understands what switching solutions would require.

Join ProcureCon IT Sourcing to compare the questions procurement leaders ask AI suppliers and how they verify the controls their organizations need.