ML Security PapersDETree: DEtecting Human-AI Collaborative Texts via Tree-Structured Hierarchical Representation Learning
Yongxin He 1,2,3, Shan Zhang 4,3, Yixuan Cao 1,2,3, Lei Ma 4,3, Ping Luo 1,2,3
1 Institute of Computing Technology, Chinese Academy of Sciences
Published on arXiv
2510.17489
Output Integrity Attack
OWASP ML Top 10 — ML09
Key Finding
DETree improves hybrid text detection performance and significantly enhances OOD robustness, particularly in few-shot learning conditions, over binary and multi-class baselines
DETree (Hierarchical Affinity Tree)
Novel technique introduced
Detecting AI-involved text is essential for combating misinformation, plagiarism, and academic misconduct. However, AI text generation includes diverse collaborative processes (AI-written text edited by humans, human-written text edited by AI, and AI-generated text refined by other AI), where various or even new LLMs could be involved. Texts generated through these varied processes exhibit complex characteristics, presenting significant challenges for detection. Current methods model these processes rather crudely, primarily employing binary classification (purely human vs. AI-involved) or multi-classification (treating human-AI collaboration as a new class). We observe that representations of texts generated through different processes exhibit inherent clustering relationships. Therefore, we propose DETree, a novel approach that models the relationships among different processes as a Hierarchical Affinity Tree structure, and introduces a specialized loss function that aligns text representations with this tree. To facilitate this learning, we developed RealBench, a comprehensive benchmark dataset that automatically incorporates a wide spectrum of hybrid texts produced through various human-AI collaboration processes. Our method improves performance in hybrid text detection tasks and significantly enhances robustness and generalization in out-of-distribution scenarios, particularly in few-shot learning conditions, further demonstrating the promise of training-based approaches in OOD settings. Our code and dataset are available at https://github.com/heyongxin233/DETree.
Key Contributions
- DETree: a hierarchical affinity tree structure that models relationships among different human-AI collaborative text generation processes, with a specialized alignment loss function
- RealBench: a comprehensive benchmark dataset automatically covering a wide spectrum of hybrid texts from AI-written+human-edited, human-written+AI-edited, and AI-refined-by-AI processes
- Demonstrated improved robustness and generalization in out-of-distribution settings, including few-shot learning scenarios
🛡️ Threat Analysis
Proposes a novel AI-generated content detection architecture (DETree with Hierarchical Affinity Tree structure) specifically targeting detection of LLM-involved text across diverse human-AI collaborative generation processes — a direct ML09 contribution to output integrity and content authenticity.