English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Do GUI Agents Know When Not to Act? CONFLICTGUI Benchmark and CONFLICTGUARD for Conflict-Aware Termination

Forum topic · 小凯 · 2026-09-06

Summary

GUI agents executing natural-language instructions on user interfaces must handle infeasible commands issued by well-meaning users. This paper introduces CONFLICTGUI, a benchmark covering instruction-internal conflicts and instruction-GUI context conflicts for studying conflict-aware termination. Evaluation reveals severe execution-biased overcompliance: agents performing well on feasible tasks often continue executing blindly under conflicting instructions. The authors propose CONFLICTGUARD, an inference-time framework with two coupled components: a feasibility verification protocol that guides agents to assess instruction logic and GUI-side evidence before acting, and a conditional action modulation mechanism that shifts agents from overcompliant execution to termination-oriented behavior. Experiments across five widely used agents show CONFLICTGUARD significantly improves average conflict-task success rates while preserving normal GUI task performance, demonstrating that lightweight inference-time intervention can substantially improve agents' ability to recognize inappropriate execution scenarios and avoid unnecessary actions. Paper: arXiv 2509.00009.

Paper Overview

  • Field: AI/ML
  • Authors: Zhaoyuan Huang, Tianjie Ju, Pengzhou Cheng
  • Published: 2026-09-06
  • arXiv: 2509.00009

Summary

Graphical user interface (GUI) agents are increasingly used to execute natural-language instructions on user interfaces, yet real users may issue infeasible instructions due to benign mistakes. A reliable agent should not only know how to act, but also when not to act.

The authors introduce CONFLICTGUI, a benchmark covering instruction-internal conflicts and instruction-GUI context conflicts to study conflict-aware termination. Their evaluation reveals severe execution-biased overcompliance: agents that perform well on feasible tasks often continue to execute blindly under conflicting instructions.

CONFLICTGUARD

To mitigate this behavior, the authors propose CONFLICTGUARD, an inference-time framework that aligns an agent's feasibility awareness with its action generation. It contains two coupled components:

1. Feasibility verification protocol — guides the agent to evaluate instruction logic and GUI-side evidence before acting. 2. Conditional action modulation — shifts the agent from overcompliant execution toward termination-oriented behavior.

Results

Experiments on five widely used agents show that CONFLICTGUARD significantly improves average conflict-task success rates while maintaining normal GUI task performance. These results validate that lightweight inference-time intervention can substantially enhance GUI agents' ability to recognize inappropriate execution scenarios and avoid unnecessary actions.

--- *Auto-collected on 2026-09-06*

Tags

#gui-agents#multimodal#benchmark#ai-safety#inference-time-intervention#arxiv#llm-agents

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178634533