PINA: Prompt Injection Attack against Navigation 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_ | 1866917211310915584 |
|---|---|
| author | Liu, Jiani He, Yixin Fan, Lanlan Zhong, Qidi Cheng, Yushi Zhang, Meng Chen, Yanjiao Xu, Wenyuan |
| author_facet | Liu, Jiani He, Yixin Fan, Lanlan Zhong, Qidi Cheng, Yushi Zhang, Meng Chen, Yanjiao Xu, Wenyuan |
| contents | Navigation agents powered by large language models (LLMs) convert natural language instructions into executable plans and actions. Compared to text-based applications, their security is far more critical: a successful prompt injection attack does not just alter outputs but can directly misguide physical navigation, leading to unsafe routes, mission failure, or real-world harm. Despite this high-stakes setting, the vulnerability of navigation agents to prompt injection remains largely unexplored. In this paper, we propose PINA, an adaptive prompt optimization framework tailored to navigation agents under black-box, long-context, and action-executable constraints. Experiments on indoor and outdoor navigation agents show that PINA achieves high attack success rates with an average ASR of 87.5%, surpasses all baselines, and remains robust under ablation and adaptive-attack conditions. This work provides the first systematic investigation of prompt injection attacks in navigation and highlights their urgent security implications for embodied LLM agents. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_13612 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | PINA: Prompt Injection Attack against Navigation Agents Liu, Jiani He, Yixin Fan, Lanlan Zhong, Qidi Cheng, Yushi Zhang, Meng Chen, Yanjiao Xu, Wenyuan Cryptography and Security Navigation agents powered by large language models (LLMs) convert natural language instructions into executable plans and actions. Compared to text-based applications, their security is far more critical: a successful prompt injection attack does not just alter outputs but can directly misguide physical navigation, leading to unsafe routes, mission failure, or real-world harm. Despite this high-stakes setting, the vulnerability of navigation agents to prompt injection remains largely unexplored. In this paper, we propose PINA, an adaptive prompt optimization framework tailored to navigation agents under black-box, long-context, and action-executable constraints. Experiments on indoor and outdoor navigation agents show that PINA achieves high attack success rates with an average ASR of 87.5%, surpasses all baselines, and remains robust under ablation and adaptive-attack conditions. This work provides the first systematic investigation of prompt injection attacks in navigation and highlights their urgent security implications for embodied LLM agents. |
| title | PINA: Prompt Injection Attack against Navigation Agents |
| topic | Cryptography and Security |
| url | https://arxiv.org/abs/2601.13612 |