# When the Forger Is the Judge: GPT-Image-2 Cannot Recognize Its Own Faked Documents

arXiv: https://arxiv.org/abs/2604.25213 · HTML: https://mlsanity.com/paper/2604.25213/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 13 of its 24 references that are in the map.
- First posted on arXiv 28 Apr 2026.

## Builds on (most cited first, 13 of 13)

- [LLM Evaluators Recognize and Favor Their Own Generations](https://mlsanity.com/paper/2404.13076/) (2024)
- [Hierarchical Fine-Grained Image Forgery Detection and Localization](https://mlsanity.com/paper/2303.17111/) (2023)
- [PSCC-Net: Progressive Spatio-Channel Correlation Network for Image Manipulation Detection and Localization](https://mlsanity.com/paper/2103.10596/) (2021)
- [TruFor: Leveraging all-round clues for trustworthy image forgery detection and localization](https://mlsanity.com/paper/2212.10957/) (2022)
- [AEROBLADE: Training-Free Detection of Latent Diffusion Images Using Autoencoder Reconstruction Error](https://mlsanity.com/paper/2401.17879/) (2024)
- [IML-ViT: Benchmarking Image Manipulation Localization by Vision Transformer](https://mlsanity.com/paper/2307.14863/) (2023)
- [Spatial Dual-Modality Graph Reasoning for Key Information Extraction](https://mlsanity.com/paper/2103.14470/) (2021)
- [Revisiting Tampered Scene Text Detection in the Era of Generative AI](https://mlsanity.com/paper/2407.21422/) (2024)
- [Toward Real Text Manipulation Detection: New Dataset and New Solution](https://mlsanity.com/paper/2312.06934/) (2023)
- [Can Multi-modal (reasoning) LLMs work as deepfake detectors?](https://mlsanity.com/paper/2503.20084/) (2025)
- [SAGI: Semantically Aligned and Uncertainty Guided AI Image Inpainting](https://mlsanity.com/paper/2502.06593/) (2025)
- [Can Multi-modal (reasoning) LLMs detect document manipulation?](https://mlsanity.com/paper/2508.11021/) (2025)
- [AIForge-Doc: A Benchmark for Detecting AI-Forged Tampering in Financial and Form Documents](https://mlsanity.com/paper/2602.20569/) (2026)

## Built on by (most cited first, 3 of 3)

- [ChatGPT Images 2.5 on Forgery Tasks: Testing Advertised Improvements Against Known Answers](https://mlsanity.com/paper/2609.13617/) (2026)
- [SynCred-Bench: Benchmarking Synthetic Credibility in AI-Generated Visual Misinformation](https://mlsanity.com/paper/2606.03348/) (2026)
- [Agents That Edit Documents: Measuring Agentic PDF Forgery Against a Non-Agentic Control](https://mlsanity.com/paper/2609.23953/) (2026)

## Abstract (arXiv)

OpenAI's GPT-Image-2 has effectively erased the visual boundary between authentic and AI-edited document images: a single number on a receipt can be replaced in under a second for a few cents. We release AIForge-Doc v2, a paired dataset of 3,066 GPT-Image-2 document forgeries with pixel-precise masks in DocTamper-compatible format, and benchmark four lines of defence: human inspectors (N=120, n=365 pair-votes via the public 2AFC site CanUSpotAI.com), TruFor (generic forensic), DocTamper (qcf-568, document-specific), and the same GPT-Image-2 model as a zero-shot self-judge -- asked, to avoid the trivial "image is mostly real" reading, whether any region was generated or edited by an AI image model. Human 2AFC accuracy is 0.501, indistinguishable from chance: even side-by-side, inspectors cannot tell GPT-Image-2 receipt forgeries from authentic counterparts. The three computational judges sit only modestly above (TruFor 0.599, DocTamper 0.585, self-judge 0.532). The self-judge fails consistently, not by chance: across five prompt strategies and four policies for handling ambiguous responses, AUC never rises above 0.59. To rule out the possibility that the two forensic detectors are broken on our source domain rather than blind to AI inpainting, we calibrate each on a same-domain traditional-tampering set built for its training distribution: TruFor reaches AUC 0.962 on cross-camera splicing of our dataset, DocTamper reaches 0.852 on cross-document OCR-token splicing with two-pass JPEG re-encoding. Both retain near-published performance on traditional tampering; switching to GPT-Image-2 inpainting drops AUC by 0.27-0.36 (0.962->0.599 TruFor; 0.852->0.585 DocTamper), isolating a detection gap specific to GPT-Image-2 inpainting. We release the dataset, pipeline, four-judge protocol, and calibration sets.