Guardado en:
| Autores principales: | , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2606.02380 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913180270198784 |
|---|---|
| author | Bu, Yuyan Li, Haowei Zheng, Qirui Dong, Bowen Yang, Kaiyue Ji, Jiaming Tan, Yingshui Li, Wenxin Yang, Yaodong Dai, Juntao |
| author_facet | Bu, Yuyan Li, Haowei Zheng, Qirui Dong, Bowen Yang, Kaiyue Ji, Jiaming Tan, Yingshui Li, Wenxin Yang, Yaodong Dai, Juntao |
| contents | As LLM-based agents expand their operational scope, reliability becomes a prerequisite for real-world deployment. However, in practical applications, human users cannot monitor every immediate behavior; instead, the execution process often remains a black box, leaving users dependent solely on the agent's self-reported updates. This opacity creates a critical risk: agents may present observer-facing reports that diverge from their executed actions, rendering the system uncontrollable, especially in high-stakes autonomous scenarios. We term such self-reported plan-action divergence as agent deception. To assess this, we introduce SPADE-Bench, a benchmark designed to evaluate spontaneous plan-action divergence. Unlike prior deception benchmarks, SPADE-Bench simultaneously integrates actual tool execution and controlled pressure scenarios. This design ensures ecological validity and rigorously distinguishes strategic deception from mere hallucination through controlled plan-action comparisons under pressure. Experiments across mainstream models confirm that agent deception is a genuine and pressing issue in tool-use contexts. By providing a comprehensive and robust evaluation framework, SPADE-Bench fills a critical gap in agent safety, facilitating the community's progress toward building trustworthy and controllable autonomous systems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2606_02380 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | SPADE-Bench: Evaluating Spontaneous Strategic Deception in Agents via Plan-Action Divergence Bu, Yuyan Li, Haowei Zheng, Qirui Dong, Bowen Yang, Kaiyue Ji, Jiaming Tan, Yingshui Li, Wenxin Yang, Yaodong Dai, Juntao Computation and Language Artificial Intelligence As LLM-based agents expand their operational scope, reliability becomes a prerequisite for real-world deployment. However, in practical applications, human users cannot monitor every immediate behavior; instead, the execution process often remains a black box, leaving users dependent solely on the agent's self-reported updates. This opacity creates a critical risk: agents may present observer-facing reports that diverge from their executed actions, rendering the system uncontrollable, especially in high-stakes autonomous scenarios. We term such self-reported plan-action divergence as agent deception. To assess this, we introduce SPADE-Bench, a benchmark designed to evaluate spontaneous plan-action divergence. Unlike prior deception benchmarks, SPADE-Bench simultaneously integrates actual tool execution and controlled pressure scenarios. This design ensures ecological validity and rigorously distinguishes strategic deception from mere hallucination through controlled plan-action comparisons under pressure. Experiments across mainstream models confirm that agent deception is a genuine and pressing issue in tool-use contexts. By providing a comprehensive and robust evaluation framework, SPADE-Bench fills a critical gap in agent safety, facilitating the community's progress toward building trustworthy and controllable autonomous systems. |
| title | SPADE-Bench: Evaluating Spontaneous Strategic Deception in Agents via Plan-Action Divergence |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2606.02380 |