GAMMA: Generalizable Alignment via Multi-task and Manipulation-Augmented Training for AI-Generated Image Detection

2025-09-12

AI Summary

Current AI-generated image detectors perform well on in-distribution data but exhibit poor generalization to unseen generative models, primarily due to over-reliance on model-specific artifacts (e.g., style priors, compression traces). To address this cross-domain generalization challenge, we propose a multi-task detection framework incorporating operation consistency constraints and a reverse cross-attention mechanism—enabling the segmentation head to guide the classification branch in correcting domain bias. Our method further integrates inpainting-based operation augmentation, semantics-preserving perturbations, and joint supervision from dual segmentation heads and a classification head. Evaluated on the GenImage benchmark, our approach achieves state-of-the-art generalization performance, improving accuracy by 5.8% over prior methods, and demonstrates significant robustness against emerging generative models such as GPT-4o.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Natural Language Processing: Generation
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

With generative models becoming increasingly sophisticated and diverse, detecting AI-generated images has become increasingly challenging. While existing AI-genereted Image detectors achieve promising performance on in-distribution generated images, their generalization to unseen generative models remains limited. This limitation is largely attributed to their reliance on generation-specific artifacts, such as stylistic priors and compression patterns. To address these limitations, we propose GAMMA, a novel training framework designed to reduce domain bias and enhance semantic alignment. GAMMA introduces diverse manipulation strategies, such as inpainting-based manipulation and semantics-preserving perturbations, to ensure consistency between manipulated and authentic content. We employ multi-task supervision with dual segmentation heads and a classification head, enabling pixel-level source attribution across diverse generative domains. In addition, a reverse cross-attention mechanism is introduced to allow the segmentation heads to guide and correct biased representations in the classification branch. Our method achieves state-of-the-art generalization performance on the GenImage benchmark, imporving accuracy by 5.8%, but also maintains strong robustness on newly released generative model such as GPT-4o.

Problem

• Detecting AI-generated images across unseen generative models
• Reducing reliance on generation-specific artifacts and biases
• Enhancing generalization and robustness in image detection

Innovation

• Multi-task supervision with dual segmentation
• Diverse manipulation strategies for consistency
• Reverse cross-attention mechanism correcting bias