最近一年的 AI 安全论文
本站共收录 65 篇- 思维链监督需为忠实性与可监控性分别提供证据
Chain-of-Thought Oversight Requires Separate Evidence for Faithfulness and Monitorability
- FAME：用反事实推理评测自主智能体的虚假记忆
Beyond the Shadows of Plato's Cave: Evaluating False Memory in Autonomous Agents via Counterfactual Reasoning
- TAME：用 SAE 抑制激活缓解 VLM 思维链混淆
Taming CoT Obfuscation in VLMs: From Mechanistic Evidence to Activation Enforcement
- 人类对齐的模型是人类行为的模型吗？偏好对齐中的图灵测试差距
Are Human-Aligned Models Models of Humans? A Turing-Test Gap in Preference Alignment
- AntiSkillBench：评测 persona skill 的隐私泄露与行为模仿风险
When Agents Learn to Be You: Benchmarking Privacy Leakage, Impersonation Risk, and Defenses in Persona Skills
- 生成图像更易被遗忘：面向合成图像检测的机器遗忘视角
Generated Images Are Easier to Forget: A Machine Unlearning Perspective for Synthetic Image Detection
- TC-UAP：针对图像转视频与微调定制的通用视频保护方法
Delving into the Temporal Challenges of Unified Video Protection Against Image-to-Video and Fine-Tuning-based Customization
- MLUBench：面向多模态大语言模型持续遗忘评测的基准
MLUBench: A Benchmark for Lifelong Unlearning Evaluation in MLLMs
- 面向多模态预训练模型的可调文本引导后门攻击（自然词触发）
Adjustable Text-Guided Backdoor Attacks with Natural-Word Triggers on Multimodal Pretrained Models
- GaussLock 用属性空间陷阱为 3D Gaussian 生成模型抵御未授权微调
Immunizing 3D Gaussian Generative Models Against Unauthorized Fine-Tuning via Attribute-Space Traps
- VII：面向图像到视频生成模型越狱的视觉指令注入
VII: Visual Instruction Injection for Jailbreaking Image-to-Video Generation Models
- 当安全相冲突：用自适应安全引导解决文本到图像扩散模型的多类别有害冲突
When Safety Collides: Resolving Multi-Category Harmful Conflicts in Text-to-Image Diffusion via Adaptive Safety Guidance
- 梯度上升真的必要吗？机器遗忘的“先记忆后遗忘”方法
Is Gradient Ascent Really Necessary? Memorize to Forget for Machine Unlearning
- 隐私保护模型转录：基于差分隐私合成蒸馏
Privacy-Preserving Model Transcription with Differentially Private Synthetic Distillation
- AdLift：把 2D 对抗扰动提升为 3D Gaussian Splatting 资产防护，抵御指令驱动编辑
AdLift: Lifting Adversarial Perturbations to Safeguard 3D Gaussian Splatting Assets Against Instruction-Driven Editing
- RDSplat：抗 2D 与 3D 扩散编辑的 3D Gaussian Splatting 鲁棒水印框架
RDSplat: Robust Watermarking for 3D Gaussian Splatting Against 2D and 3D Diffusion Editing
- DINO-Detect：一个简洁有效的抗模糊 AI 生成图像检测框架
DINO-Detect: A Simple yet Effective Framework for Blur-Robust AI-Generated Image Detection
署名以 dblp 和本人的 Google Scholar 为准；只列本站收录的 AI 安全论文。