---
title: "Generalizable AI-Generated Image Detection with Embedding-Space Explanations"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2611.04111v1"
permalink: "https://feed.astralor.com/items/itm_73750b4e96aa3b3b4125a1e7"
published: "2026-11-03T00:38:54.000Z"
score: 57
channel: "ai"
tags: ["多模态","Generalizable AI-Generated Image","Detection","Embedding-Space Explanations AI-generated","Most","AI-generated","Specialist","DINOv3"]
---

# Generalizable AI-Generated Image Detection with Embedding-Space Explanations

## 速览

AI-generated images have reached unprecedented realism, while instruction-driven AI editors re-synthesize images for local changes, leaving no splice boundary to localize. Most detectors are black boxes with unverifiabl…

## 关键点

- AI-generated images have reached unprecedented realism, while instruction-driven AI editors re-synthesize images for local changes, leaving no splice boundary to localize.
- Most detectors are black boxes with unverifiable verdicts and generalize poorly because adapting a backbone on the detection distribution memorizes generator-specific fingerprints that quickly become obsolete.
- Specialist detectors built on pure-vision representations, particularly DINOv3, generalize well across generators, but explainable methods forgo this advantage: jointly training one multimodal model for classification a…

## 全文

AI-generated images have reached unprecedented realism, while instruction-driven AI editors re-synthesize images for local changes, leaving no splice boundary to localize. Most detectors are black boxes with unverifiable verdicts and generalize poorly because adapting a backbone on the detection distribution memorizes generator-specific fingerprints that quickly become obsolete. Specialist detectors built on pure-vision representations, particularly DINOv3, generalize well across generators, but explainable methods forgo this advantage: jointly training one multimodal model for classification and language generation subjects its visual representation to the language objective. We therefore require that explanation training cannot alter the verdict. We introduce FOX, which first builds and locks a specialist DINOv3 detector, then uses its representation to guide a lightweight VLM toward forensic explanation. A joint-embedding predictor maps locked visual evidence into a frozen text encoder's semantic space, producing guidance that the VLM verbalizes on demand. FOX achieves 97.4 mean AUC across 14 unseen generators, 69.8 BERTScore on XPlainVerse, and 56.1 on LOKI.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_73750b4e96aa3b3b4125a1e7