From Evidence to Verdict: An Agent-Based Forensic Framework for AI-Generated Image Detection

TL;DR

The paper presents AIFo, a training-free, agent-based forensic framework for AI-generated image detection. By coordinating a toolbox of external forensic tools through specialized LLM-based agents and a structured debate mechanism, AIFo achieves high accuracy and robustness across lab and in-the-wild data, outperforming traditional classifiers and single VLMs. A memory-augmented reasoning module further enhances performance by leveraging historical cases to refine evidence weighting and recover misclassifications. The work demonstrates a shift from static detectors to interpretable, procedural reasoning that can adapt to rapid evolution in generative models, offering a scalable approach for information integrity and media authenticity.

Abstract

The rapid evolution of AI-generated images poses unprecedented challenges to information integrity and media authenticity. Existing detection approaches suffer from fundamental limitations: traditional classifiers lack interpretability and fail to generalize across evolving generative models, while vision-language models (VLMs), despite their promise, remain constrained to single-shot analysis and pixel-level reasoning. To address these challenges, we introduce AIFo (Agent-based Image Forensics), a novel training-free framework that emulates human forensic investigation through multi-agent collaboration. Unlike conventional methods, our framework employs a set of forensic tools, including reverse image search, metadata extraction, pre-trained classifiers, and VLM analysis, coordinated by specialized LLM-based agents that collect, synthesize, and reason over cross-source evidence. When evidence is conflicting or insufficient, a structured multi-agent debate mechanism allows agents to exchange arguments and reach a reliable conclusion. Furthermore, we enhance the framework with a memory-augmented reasoning module that learns from historical cases to improve future detection accuracy. Our comprehensive evaluation spans 6,000 images across both controlled laboratory settings and challenging real-world scenarios, including images from modern generative platforms and diverse online sources. AIFo achieves 97.05% accuracy, substantially outperforming traditional classifiers and state-of-the-art VLMs. These results demonstrate that agent-based procedural reasoning offers a new paradigm for more robust, interpretable, and adaptable AI-generated image detection.

Figures (6)

• Figure 1: High-level overview of our proposed AIFo.
• Figure 2: Examples of our agent framework's decision-making process, demonstrating diverse evidence integration across different image types and sources.
• Figure 3: Analysis of individual tool contributions to the agent framework: (a) reliability rates measuring agent trust in each tool's evidence, and (b) coverage rates showing the proportion of decisions where each tool provides informative evidence.
• Figure 4: Performance degradation when each tool is disabled from the framework.
• Figure 5: Overview of the memory-augmented reasoning module.