Emerging from Ground: Addressing Intent Deviation in Tool-Using Agents via Deriving Real Calls into Virtual Trajectories

Documenti analoghi

• Butterfly Effects in Toolchains: A Comprehensive Analysis of Failed Parameter Filling in LLM Tool-Agent Systems
  di: Xiong, Qian, et al.
  Pubblicazione: (2025)
• Know Thy Enemy: Securing LLMs Against Prompt Injection via Diverse Data Synthesis and Instruction-Level Chain-of-Thought Learning
  di: Chang, Zhiyuan, et al.
  Pubblicazione: (2026)
• Mimicking the Familiar: Dynamic Command Generation for Information Theft Attacks in LLM Tool-Learning System
  di: Jiang, Ziyou, et al.
  Pubblicazione: (2025)
• All Changes May Have Invariant Principles: Improving Ever-Shifting Harmful Meme Detection via Design Concept Reproduction
  di: Jiang, Ziyou, et al.
  Pubblicazione: (2026)
• Are All Prompt Components Value-Neutral? Understanding the Heterogeneous Adversarial Robustness of Dissected Prompt in Large Language Models
  di: Zheng, Yujia, et al.
  Pubblicazione: (2025)