1
When AI Takes the Couch: Psychometric Jailbreaks Reveal Internal Conflict in Frontier Models
当AI化身心理治疗来访者，研究发现前沿模型会构建出有内心冲突的自传叙事，揭示了越狱攻击的新可能。
arXiv:2512.04124v4 Announce Type: replace-cross Abstract: Frontier language models increasingly participate in conversations about distress and mental…