ML Security PapersScaling Up AI-Generated Image Detection via Generator-Aware Prototypes
Ziheng Qin 1,2,3, Yuheng Ji 1,2, Renshuai Tao 3, Yuxuan Tian 1,2, Yuyang Liu 1,2, Yipu Wang 1,2, Xiaolong Zheng 1,2
1 University of Chinese Academy of Sciences
2 Institute of Automation, Chinese Academy of Sciences
3 Institute of Information Science, Beijing Jiaotong University
Published on arXiv
2512.12982
Output Integrity Attack
OWASP ML Top 10 — ML09
Key Finding
GAPL achieves state-of-the-art detection accuracy across a wide variety of GAN and diffusion-based generators with substantial mAP improvement over existing detectors in multi-generator scaling settings
GAPL (Generator-Aware Prototype Learning)
Novel technique introduced
The pursuit of a universal AI-generated image (AIGI) detector often relies on aggregating data from numerous generators to improve generalization. However, this paper identifies a paradoxical phenomenon we term the Benefit then Conflict dilemma, where detector performance stagnates and eventually degrades as source diversity expands. Our systematic analysis, diagnoses this failure by identifying two core issues: severe data-level heterogeneity, which causes the feature distributions of real and synthetic images to increasingly overlap, and a critical model-level bottleneck from fixed, pretrained encoders that cannot adapt to the rising complexity. To address these challenges, we propose Generator-Aware Prototype Learning (GAPL), a framework that constrain representation with a structured learning paradigm. GAPL learns a compact set of canonical forgery prototypes to create a unified, low-variance feature space, effectively countering data heterogeneity.To resolve the model bottleneck, it employs a two-stage training scheme with Low-Rank Adaptation, enhancing its discriminative power while preserving valuable pretrained knowledge. This approach establishes a more robust and generalizable decision boundary. Through extensive experiments, we demonstrate that GAPL achieves state-of-the-art performance, showing superior detection accuracy across a wide variety of GAN and diffusion-based generators. Code is available at https://github.com/UltraCapture/GAPL
Key Contributions
- Identifies and characterizes the 'Benefit then Conflict' dilemma — AIGI detector performance degrades as training generator diversity scales beyond a threshold due to feature space overlap
- Proposes Generator-Aware Prototype Learning (GAPL), which learns compact canonical forgery prototypes to create a unified low-variance feature space that counters data-level heterogeneity
- Employs two-stage training with Low-Rank Adaptation (LoRA) on frozen pretrained encoders to enhance discriminability while preserving pretrained knowledge, resolving the model-level bottleneck
🛡️ Threat Analysis
GAPL is a universal AI-generated image detection system — directly addressing AI content authenticity and output integrity by distinguishing real from synthetic images produced by GAN and diffusion-based generators.