Interpretable and Reliable Detection of AI-Generated Images via Grounded Reasoning in MLLMs

2025-06-08

AI Summary

Existing AI-generated image detection methods achieve high accuracy but lack interpretability, while multimodal large language models (MLLMs) often suffer from visual hallucinations in forgery detection, yielding explanations inconsistent with image content and human reasoning. To address this, we propose a vision–language joint reasoning framework that unifies detection, artifact localization, and natural-language explanation for the first time. Our key contributions are: (1) constructing the first AI-image annotation dataset featuring bounding-box-level artifact localization and fine-grained textual defect descriptions; (2) designing a multi-stage collaborative supervised fine-tuning strategy integrating visual grounding reasoning, artifact localization learning, and explanation generation; and (3) achieving state-of-the-art performance across detection accuracy, localization precision, and explanation consistency—significantly enhancing model trustworthiness and human interpretability.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Language and Vision
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

The rapid advancement of image generation technologies intensifies the demand for interpretable and robust detection methods. Although existing approaches often attain high accuracy, they typically operate as black boxes without providing human-understandable justifications. Multi-modal Large Language Models (MLLMs), while not originally intended for forgery detection, exhibit strong analytical and reasoning capabilities. When properly fine-tuned, they can effectively identify AI-generated images and offer meaningful explanations. However, existing MLLMs still struggle with hallucination and often fail to align their visual interpretations with actual image content and human reasoning. To bridge this gap, we construct a dataset of AI-generated images annotated with bounding boxes and descriptive captions that highlight synthesis artifacts, establishing a foundation for human-aligned visual-textual grounded reasoning. We then finetune MLLMs through a multi-stage optimization strategy that progressively balances the objectives of accurate detection, visual localization, and coherent textual explanation. The resulting model achieves superior performance in both detecting AI-generated images and localizing visual flaws, significantly outperforming baseline methods.

Problem

• Detect AI-generated images interpretably and reliably
• Align MLLMs' visual interpretations with human reasoning
• Improve detection accuracy and provide meaningful explanations

Innovation

• Use MLLMs for AI-generated image detection
• Fine-tune with multi-stage optimization strategy
• Annotate dataset for visual-textual grounded reasoning