WildAIGC-Bench: Can AI Image Detectors Handle the Wild? A Benchmark of Generated and Edited Images in Online Circulation

Key points

• AI-generated images linked to news events and online claims can mislead audiences, enable fraud, or cause public controversy.1
• The benchmark contains 407 verified AI-generated or edited images, each accompanied by a source webpage link and an evidence summary.1
• Evaluation of dedicated visual detectors, multimodal large language models, and a simple Perception Encoder baseline reveals substantial recall gaps relative to existing benchmarks.1
• WildAIGC-Bench provides both a practical detection challenge and a basis for continued data expansion and source-based verification.1

Sources (1)

• [1]WildAIGC-Bench: Can AI Image Detectors Handle the Wild? A Benchmark of Generated and Edited Images in Online Circulation
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 20, 12:07 PM

  We introduce WildAIGC-Bench, built through recurring collection across news websites, fact-checking sources, and related social media content.

  AI-generated images linked to news events and online claims can mislead audiences, enable fraud, or cause public controversy.

Extractive summary: sentences quoted from the sources.