The chatbot that invented a refund policy
In a widely reported 2024 case, an airline’s support chatbot told a customer about a bereavement-fare refund policy that did not exist. A tribunal held the airline responsible for its AI’s statement. The output was confident, fluent, and completed — and wrong.
Missing layer: output verification against an authoritative policy source before the answer reached the customer.
Would have caught it: a VerifyAPI gate checking the claim against the published policy, recording pass/fail.
Lesson: “the AI said it” is not a defense. Completed is not correct.
Summary of a publicly reported incident; see primary reporting for full detail.

