OmniAID: Decoupling Semantic and Artifacts for Universal AI-Generated Image Detection in the Wild

2025-11-11

AI Summary

Current AI-generated image (AIGI) detectors suffer from poor generalization and conflate content-related artifacts (e.g., semantic inconsistencies) with content-agnostic distortions (e.g., high-frequency noise). To address this, we propose a decoupled universal detection framework. Methodologically, we design a Mixture-of-Experts (MoE) architecture: a fixed universal artifact expert captures cross-model, content-agnostic distortions, while routable domain-specific experts model semantic-aware generation errors; training employs hard sampling and a lightweight gating network in two stages. Our key contribution is the first explicit disentanglement of these two defect categories, significantly improving robustness across diverse generative models and semantic content. Evaluated on standard benchmarks and the newly constructed large-scale photorealistic dataset Mirage, our method achieves state-of-the-art performance, reducing false positive rates by 12.7% in complex scenarios.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Mixture of Experts (MoE)
• Philosophy and Ethics of AI: Artificial General Intelligence

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

A truly universal AI-Generated Image (AIGI) detector must simultaneously generalize across diverse generative models and varied semantic content. Current state-of-the-art methods learn a single, entangled forgery representation--conflating content-dependent flaws with content-agnostic artifacts--and are further constrained by outdated benchmarks. To overcome these limitations, we propose OmniAID, a novel framework centered on a decoupled Mixture-of-Experts (MoE) architecture. The core of our method is a hybrid expert system engineered to decouple: (1) semantic flaws across distinct content domains, and (2) these content-dependent flaws from content-agnostic universal artifacts. This system employs a set of Routable Specialized Semantic Experts, each for a distinct domain (e.g., human, animal), complemented by a Fixed Universal Artifact Expert. This architecture is trained using a bespoke two-stage strategy: we first train the experts independently with domain-specific hard-sampling to ensure specialization, and subsequently train a lightweight gating network for effective input routing. By explicitly decoupling"what is generated"(content-specific flaws) from"how it is generated"(universal artifacts), OmniAID achieves robust generalization. To address outdated benchmarks and validate real-world applicability, we introduce Mirage, a new large-scale, contemporary dataset. Extensive experiments, using both traditional benchmarks and our Mirage dataset, demonstrate our model surpasses existing monolithic detectors, establishing a new, robust standard for AIGI authentication against modern, in-the-wild threats.

Problem

• Decouples semantic flaws from universal artifacts for AI image detection
• Generalizes across diverse generative models and varied semantic content
• Overcomes limitations of entangled representations and outdated benchmarks

Innovation

• Decouples semantic flaws from universal artifacts
• Employs routable specialized and fixed universal experts
• Uses two-stage training with domain-specific hard-sampling