AI hallucination

Hallucinations are reduced by grounding answers in approved sources, requiring evidence before claiming an action succeeded, and letting the AI say "I don't know." Why it happens: language models generate likely-sounding text. When they lack the right information, they can fill the gap with something that sounds plausible but is wrong, like an invented return policy or a tracking number that does not exist. In customer service, that turns into broken promises. Checking claims against real records before they reach the customer is the most reliable defense.