AEGIS: A Holistic Benchmark for Evaluating Forensic Analysis of AI-Generated Academic Images

TL;DR

AEGIS is a comprehensive benchmark that evaluates the forensic analysis of AI-generated academic images across multiple dimensions, revealing current limitations and strengths in detection, localization, and reasoning.

Contribution

It introduces a novel, multi-faceted benchmark covering diverse academic categories, forgery strategies, and evaluation metrics, advancing the assessment of AI-generated image forensics.

Findings

GPT-5.1 reaches 48.80% overall performance in forensic detection.

Most forgery methods have average accuracy below 50%.

Multimodal large language models achieve 84.74% accuracy in textual artifact recognition.

Abstract

We introduce AEGIS, A holistic benchmark for Evaluating forensic analysis of AI-Generated academic ImageS. Compared to existing benchmarks, AEGIS features three key advances: (1) Domain-Specific Complexity: covering seven academic categories with 39 fine-grained subtypes, exposing intrinsic forensic difficulty, where even GPT-5.1 reaches 48.80% overall performance and expert models achieve only limited localization accuracy (IoU 30.09%); (2) Diverse Forgery Simulations: modeling four prevalent academic forgery strategies across 25 generative models, with 11 yielding average forensic accuracy below 50%, showing that forensics lag behind generative advances; and (3) Multi-Dimensional Forensic Evaluation: jointly assessing detection, reasoning, and localization, revealing complementary strengths between model families, with multimodal large language models (MLLMs) at 84.74% accuracy in textual artifact recognition and expert detectors peaking at 79.54% accuracy in binary authenticity detection. By evaluating 25 leading MLLMs, nine expert models, and one unified multimodal understanding and generation model, AEGIS serves as a diagnostic testbed exposing fundamental limitations in academic image forensics.