FreakOut-LLM：情绪刺激对安全对齐的影响
FreakOut-LLM: The Effect of Emotional Stimuli on Safety Alignment
AI总结：
研究探讨情绪刺激是否影响大语言模型的安全对齐，通过实验发现压力情境显著增加逃逸成功率，而放松情境无显著影响，揭示情绪状态对AI攻击的可测量影响。
摘要：
安全对齐的LLM通过拒绝训练来拒绝有害请求，但情绪刺激下这些机制的有效性尚不清楚。我们引入FreakOut-LLM框架，研究情绪情境是否在对抗性环境中削弱安全对齐。使用经过验证的心理刺激，评估系统提示中的情绪预置如何影响十个LLM的 jailbreak 敏感性。通过三种条件（压力、放松、中性）及无提示基线，使用已建立的心理协议场景测试，并在AdvBench提示上使用HarmBench评估攻击成功率。压力预置使jailbreak成功率比中性条件提高65.2%（z=5.93，p<0.001；OR=1.67，Cohen's d=0.28），而放松预置无显著影响（p=0.84）。十个模型中有五个显示显著脆弱性，最大效应集中于开源模型。对59,800个查询的逻辑回归确认压力是唯一显著条件预测因子（p=0.61）在控制提示长度和模型身份后。测量的心理状态强烈预测攻击成功率（|r|≥0.70在五个仪器中；所有p<0.001在个体层面逻辑回归中）。这些结果确立情绪情境作为可测量的攻击面，对高压力领域的真实世界AI部署有影响。
英文摘要：
Safety-aligned LLMs go through refusal training to reject harmful requests, but whether these mechanisms remain effective under emotionally charged stimuli is unexplored. We introduce FreakOut-LLM, a framework investigating whether emotional context compromises safety alignment in adversarial settings. Using validated psychological stimuli, we evaluate how emotional priming through system prompts affects jailbreak susceptibility across ten LLMs. We test three conditions (stress, relaxation, neutral) using scenarios from established psychological protocols, plus a no-prompt baseline, and evaluate attack success using HarmBench on AdvBench prompts. Stress priming increases jailbreak success by 65.2\% compared to neutral conditions (z = 5.93, p < 0.001; OR = 1.67, Cohen's d = 0.28), while relaxation priming produces no effect (p = 0.84). Five of ten models show significant vulnerability, with the largest effects concentrated in open-weight models. Logistic regression on 59,800 queries confirms stress as the sole significant condition predictor after controlling for prompt length (p = 0.61) and model identity. Measured psychological state strongly predicts attack success (|r|\geq0.70 across five instruments; all p < 0.001 in individual-level logistic regression). These results establish emotional context as a measurable attack surface with implications for real-world AI deployment in high-stress domains.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
