最近一年的 AI 安全论文
本站共收录 26 篇- Swiss-Knife：解码时可重构的外部化多目标对齐框架
Swiss-Knife: A Framework for Reconfigurable Externalised Multi-Objective Alignment at Decode Time
- MENTIS：对齐改变了什么信念？测量语言模型中的多尺度潜在扭转
MENTIS: What Belief Changes Under Alignment? Measuring Multi-Scale Latent Torsion in Language Models
- Counter Turing Test 竞赛发现：AI 生成图像检测
Findings of the Counter Turing Test: AI-Generated Image Detection
- Counter Turing Test（CT2）评测结果：AI 生成文本检测
Findings of the Counter Turing Test: AI-Generated Text Detection
- PermaFrost-Attack：以隐蔽预训练投毒（SPS）在 LLM 训练中埋设逻辑地雷
PermaFrost-Attack: Stealth Pretraining Seeding(SPS) for planting Logic Landmines During LLM Training
- CONSCIENTIA：LLM 智能体能学会运用策略吗？多智能体纽约模拟中的涌现欺骗与信任
CONSCIENTIA: Can LLM Agents Learn to Strategize? Emergent Deception and Trust in a Multi-Agent NYC Simulation
- Stochastic CHAOS：为什么确定性推理对大语言模型致命，而分布变异性是人工认知的心跳
Stochastic CHAOS: Why Deterministic Inference Kills, and Distributional Variability Is the Heartbeat of Artifical Cognition
- SPINAL：神经对齐层中的缩放定律与偏好整合
SPINAL -- Scaling-law and Preference Integration in Neural Alignment Layers
- ECLIPTICA：基于 CITA 的可切换 LLM 对齐框架
ECLIPTICA -- A Framework for Switchable LLM Alignment via CITA - Contrastive Instruction-Tuned Alignment
- AlignMerge：通过 Fisher 引导的几何约束实现保持对齐的大语言模型合并
AlignMerge - Alignment-Preserving Large Language Model Merging via Fisher-Guided Geometric Constraints
- D-STEER：在激活空间双向控制 DPO 诱导的安全行为
D-STEER: Two-Way Control of DPO-Induced Safety Behavior in Activation Space
- 对齐伪装的训练与部署不对称性：贝叶斯-Stackelberg 均衡的博弈论视角
Alignment Faking - the Train -> Deploy Asymmetry: Through a Game-Theoretic Lens with Bayesian-Stackelberg Equilibria
署名以 dblp 和本人的 Google Scholar 为准；只列本站收录的 AI 安全论文。