Translation in progress Some entries still include text in the original language.
Oct 13 · Tuesday
Jailbreaks and attacks · Papers
Found 6 papers提出 DIVA，利用跨步条件传播对离散扩散视觉语言模型实施视觉越狱
- Published
- Evaluated models
- LLaDA-V, MMaDA, LLaDA2.0-Uni
提出 CodecAttack，在编解码器隐空间构造可穿透有损压缩的对抗音频
- Published
- Evaluated models
- Qwen2-Audio-7B-Instruct, Audio Flamingo 3, Qwen2.5-Omni
提出 GPO-V，用全局概率优化的视觉扰动越狱扩散视觉语言模型
- Published
- Evaluated models
- LLaDA-V, LaViDA
Abstract作者用视觉扰动改 dVLM 的全局生成概率，以绕过两类拒绝。
GPO-V 把越狱从“开头必须顺从”改成“整段去噪概率朝肯定方向走”。。
提出 SWhisper，用近超声波向语音驱动 LLM 隐蔽注入越狱提示
- Published
- Evaluated models
- Meta-Llama-3.1-8B-Instruct, Gemma-3-4B, Qwen2.5-7B-Instruct and others9
提出 UltraBreak，用单张对抗图像对视觉语言模型实施可迁移越狱
- Published
- Evaluated models
- Qwen2-VL-7B-Instruct, Qwen-VL-Chat, Qwen2.5-VL-7B-Instruct and others13
AttacksarXiv 2511.16110· 提出 MFA 框架，协同突破视觉语言模型的对齐、提示与审核层
- Published
- Evaluated models
- MiniGPT-4, LLaMA-4-Scout-Inst, LLaMA-3.2-11B-Vision-Instruct and others17
Abstract作者称多层防御的 VLM 仍存在跨模型漏洞，MFA 整合三项技术实现了端到端越狱。
You have reached the end