最近一年的 AI 安全论文
本站共收录 52 篇- 并非你所想之处：防范推理模型对照片的地理定位，保护位置隐私
Not Where You Think: Protecting Location Privacy Against Reasoning-Based Geolocation
- AutoRedClaw：面向 Claw 风格智能体的自动化红队测试
AutoRedClaw: Automated Red-Teaming for Claw-Style Agents
- SRFT：让 Agent 从失败经验中自我反思以抵御提示注入
Self-Reflection Fine-Tuning: Enhancing Agent Security against Prompt Injection Attacks from Failure Experience
- 通过 PAC-Private 适配在扩散模型图像合成中保护敏感数据
Protecting Sensitive Data in Image Synthesis via PAC-Private Adaptation for Diffusion Models
- SAGE：基于相似度从少量已验证样本中清洗中毒训练数据
SAGE: Similarity-Based Cleaning of Poisoned Training Data from Verified Examples
- 三思而后行：通过隔离规划保护 LLM Agent 免受工具描述投毒攻击
Think Twice Before You Act: Protecting LLM Agents Against Tool Description Poisoning via Isolated Planning
- 从效率到泄露：联邦语言模型微调中的隐私后门
From Efficiency to Leakage -- Privacy Backdoor in Federated Language Model Fine-Tuning
- AgentDyn：你的智能体安全防御能否在真实动态环境中部署？
AgentDyn: Are Your Agent Security Defenses Deployable in Real-World Dynamic Environments?
- Sliced Rényi Pufferfish Privacy：方向性加性噪声机制与带梯度裁剪的私有学习
Sliced Rényi Pufferfish Privacy: Directional Additive Noise Mechanism and Private Learning with Gradient Clipping
署名以 dblp 和本人的 Google Scholar 为准；只列本站收录的 AI 安全论文。