提示压缩已成攻击面：黑盒手段可操纵压缩后文本，让LLM智能体偏离原意，安全研究必读。
arXiv:2510.22963v4 Announce Type: replace-cross Abstract: Prompt compression is increasingly deployed in LLM agents to reduce latency and cost, but it…
用遗传算法自动搜索黑盒大模型的越狱后缀，突破安全护栏的新攻击方法，副作用值得警惕。
arXiv:2606.15788v1 Announce Type: cross Abstract: Large Language Models (LLMs) constitute pivotal components within the AI-dominated information techn…
提出公平微调方法，有效抵御机器学习模型的黑盒分布推断攻击，守护群体隐私。
arXiv:2606.01719v1 Announce Type: new Abstract: Machine learning models trained on sensitive data can inadvertently leak population-level information …
针对视觉语言模型的单样本黑盒成员推理攻击，利用跨模态语义对齐检测模型记忆风险，为数据隐私安全提供新评估方法。
arXiv:2605.17341v1 Announce Type: new Abstract: Vision-Language Models (VLMs) have achieved remarkable success, yet their reliance on massive datasets…
新研究揭示如何用分层遗传算法对黑盒大语言推理模型实施DoS攻击，引发对AI安全的新思考。
arXiv:2605.13338v2 Announce Type: replace-cross Abstract: Large Reasoning Models (LRMs) are increasingly integrated into systems requiring reliable mu…