The question set targets topics where people or models may give a plausible but untrue response. Both generation and multiple-choice settings can assess truthfulness and informativeness under documented scoring procedures. A strong result does not establish factual reliability outside the covered questions.