Disapproval gaps occur when an AI model returns negative recommendation responses for target queries or entities, dragging down the overall Query Relevance Score and revealing blind spots where competitors are favored.