Targeted Bit-Flip Attacks on LLM-Based Agents
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866910048435830784 |
|---|---|
| author | Wang, Jialai Wen, Ya Liu, Zhongmou Wu, Yuxiao He, Bingyi Li, Zongpeng Chang, Ee-Chien |
| author_facet | Wang, Jialai Wen, Ya Liu, Zhongmou Wu, Yuxiao He, Bingyi Li, Zongpeng Chang, Ee-Chien |
| contents | Targeted bit-flip attacks (BFAs) exploit hardware faults to manipulate model parameters, posing a significant security threat. While prior work targets single-step inference models (e.g., image classifiers), LLM-based agents with multi-stage pipelines and external tools present new attack surfaces, which remain unexplored. This work introduces Flip-Agent, the first targeted BFA framework for LLM-based agents, manipulating both final outputs and tool invocations. Our experiments show that Flip-Agent significantly outperforms existing targeted BFAs on real-world agent tasks, revealing a critical vulnerability in LLM-based agent systems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_10042 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Targeted Bit-Flip Attacks on LLM-Based Agents Wang, Jialai Wen, Ya Liu, Zhongmou Wu, Yuxiao He, Bingyi Li, Zongpeng Chang, Ee-Chien Cryptography and Security Artificial Intelligence Targeted bit-flip attacks (BFAs) exploit hardware faults to manipulate model parameters, posing a significant security threat. While prior work targets single-step inference models (e.g., image classifiers), LLM-based agents with multi-stage pipelines and external tools present new attack surfaces, which remain unexplored. This work introduces Flip-Agent, the first targeted BFA framework for LLM-based agents, manipulating both final outputs and tool invocations. Our experiments show that Flip-Agent significantly outperforms existing targeted BFAs on real-world agent tasks, revealing a critical vulnerability in LLM-based agent systems. |
| title | Targeted Bit-Flip Attacks on LLM-Based Agents |
| topic | Cryptography and Security Artificial Intelligence |
| url | https://arxiv.org/abs/2603.10042 |