Turing Test on Screen: A Benchmark for Mobile GUI Agent Humanization

TL;DR

This paper introduces a benchmark and methods for creating mobile GUI agents that mimic human behavior convincingly to evade detection, balancing utility and human-like imitation.

Contribution

It formalizes the Turing Test on Screen as a MinMax optimization, establishes the Agent Humanization Benchmark, and proposes techniques for high imitability without losing performance.

Findings

Vanilla LMM-based agents are easily detectable due to unnatural kinematics.

Agents can achieve high imitability without sacrificing task performance.

The proposed methods improve agent humanization in adversarial environments.

Abstract

The rise of autonomous GUI agents has triggered adversarial countermeasures from digital platforms, yet existing research prioritizes utility and robustness over the critical dimension of anti-detection. We argue that for agents to survive in human-centric ecosystems, they must evolve Humanization capabilities. We introduce the ``Turing Test on Screen,'' formally modeling the interaction as a MinMax optimization problem between a detector and an agent aiming to minimize behavioral divergence. We then collect a new high-fidelity dataset of mobile touch dynamics, and conduct our analysis that vanilla LMM-based agents are easily detectable due to unnatural kinematics. Consequently, we establish the Agent Humanization Benchmark (AHB) and detection metrics to quantify the trade-off between imitability and utility. Finally, we propose methods ranging from heuristic noise to data-driven behavioral matching, demonstrating that agents can achieve high imitability theoretically and empirically without sacrificing performance. This work shifts the paradigm from whether an agent can perform a task to how it performs it within a human-centric ecosystem, laying the groundwork for seamless coexistence in adversarial digital environments.