MASH: Evading Black-Box AI-Generated Text Detectors via Style Humanization
For researchers and practitioners concerned with the reliability of AI-generated text detectors, MASH reveals a significant vulnerability that current detectors cannot withstand.
MASH is a framework that evades black-box AI-generated text detectors by humanizing style, achieving an average Attack Success Rate of 92%, outperforming baselines by 24%.
The increasing misuse of AI-generated texts (AIGT) has motivated the rapid development of AIGT detection methods. However, the reliability of these detectors remains fragile against adversarial evasions. Existing attack strategies often rely on white-box assumptions or demand prohibitively high computational and interaction costs, rendering them ineffective under practical black-box scenarios. In this paper, we propose Multi-stage Alignment for Style Humanization (MASH), a novel framework that evades black-box detectors based on style transfer. MASH sequentially employs style-injection supervised fine-tuning, direct preference optimization, and inference-time refinement to shape the distributions of AI-generated texts to resemble those of human-written texts. Experiments across 6 datasets and 5 detectors demonstrate the superior performance of MASH over 11 baseline evaders. Specifically, MASH achieves an average Attack Success Rate (ASR) of 92%, surpassing the strongest baselines by an average of 24%, while maintaining superior linguistic quality.