Errors or misalignments in AI outputs that go undetected because the user accepts the result without critical evaluation.