Rand兰德：2025年迈向人工智能评估与治理的最佳实践报告（英文版）

To measure the general capabilities and limitations of frontier artificial intelligence (AI) models and assess their societal effects and downstream risks, general-purpose AI (GPAI) model evaluations have been proposed as a valuable tool (Bengio et al., 2025). Although GPAI evaluations play an increasingly central role in institutional decisionmaking and policymaking, including the European Union (EU) AI Act’s mandate to conduct evaluations on GPAI models presenting systemic risk (European Pa