最近一年的 AI 安全论文
本站共收录 28 篇- WIFA：按意图分组监督缓解包装式越狱与过度拒答
Refusing Intent, Not Form: Wrapper-Based Intent-Group Supervision for LLM Safety
- ForesightSafety-TIDE：虚假证词使 LLM 多智能体系统事实恢复率从 72.50% 降至 14.17%
When Truth Is Distributed: Misinformation Derails Collective Fact Recovery in LLM-Based Multi-Agent Systems
- ForesightSafety-VLA：面向视觉-语言-动作模型的统一诊断安全基准
ForesightSafety-VLA: A Unified Diagnostic Safety Benchmark for Vision-Language-Action Models
- SciRisk-Bench：面向 AI4Science 安全的风险维度感知基准
SciRisk-Bench: A Risk-Dimension-Aware Benchmark for AI4Science Safety
- CogManip：评估大语言模型多轮交互中操纵行为的基准
CogManip: Benchmarking Manipulative Behavior in Multi-Turn Interactions with Large Language Model
- ForesightSafety Bench：面向安全 AI 的前沿风险评估与治理框架
ForesightSafety Bench: A Frontier Risk Evaluation and Governance Framework towards Safe AI
- Light Alignment 用单个神经元实现模型自我反思，提升大语言模型安全性
Light Alignment Improves LLM Safety via Model Self-Reflection with a Single Neuron
- 迈向脉冲神经网络对抗鲁棒性的可靠评估
Towards Reliable Evaluation of Adversarial Robustness for Spiking Neural Networks
- 《国际AI安全报告2025》第二次关键更新：技术防护与风险管理
International AI Safety Report 2025: Second Key Update: Technical Safeguards and Risk Management
署名以 dblp 和本人的 Google Scholar 为准；只列本站收录的 AI 安全论文。