arXiv cs.AI· Dennis Ng, Xingyu Shen, Ankit Raj, Kidus Zewde, Tommy Duong, Yuchen Zhou, Yuxin Zhang, Neo Tiangratanakul, Simiao Ren·· 6 小时前AI 评分48
ChatGPT Images 2.5 实测数据集发布：上线期图像收集与六款检测器评估
ChatGPT Images 2.5 in the Wild: A Launch-Period Dataset and Detector Evaluation
AI 导读
一项研究针对 ChatGPT Images 2.5 发布后图像工具底层的生成器更换导致版本归属模糊的问题，构建了覆盖 8 个来源、2,440 条帖子共 3,478 张图像的冻结数据集，发帖时间集中在公告后首 51.1 小时内。
正文 · 原文
Abstract:An image tool can change its underlying generator while retaining its public name, making version attribution from online posts ambiguous. We study this problem after the ChatGPT Images 2.5 launch. Our frozen collection contains 3,478 images from 2,440 posts across 8 sources. Recorded posting times fall within the first 51.1 hours after the announcement. It records three attribution tiers and retains standalone images after image-form filtering and targeted review. Caption claims and host records provide admission evidence, not independently verified generator identity. The observed content profile depends on the source mixture: NightCafe supplies 39.0% of images but 77.0% of CLIP-assigned fantasy scenes. We then evaluate six frozen detectors at thresholds calibrated to a 5% flag rate on reference photographs. Collection flag rates range from 3.7 to 56.4%, falling 42-81 percentage points below GenImage recall. Held-out artwork false-positive rates range from 1.5 to 96.5%, so a higher collection flag rate does not by itself establish better detection. An exploratory X-only comparison with our April collection finds a higher September flag rate for Effort, and a suggestive difference for DoU, under fixed-threshold post-clustered bootstrap intervals. Attribution, content and processing differences prevent a causal interpretation of these contrasts. The collection supports analysis of reported model use during a product transition, with source and attribution evidence retained for interpretation. The collection is released at this https URL.
| Comments: | 22 pages, 8 figures, 12 tables |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2609.15100 [cs.CV] |
| (or arXiv:2609.15100v3 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2609.15100 arXiv-issued DOI via DataCite |
Submission history
From: Simiao Ren [view email]
[v1]
Mon, 14 Sep 2026 06:25:39 UTC (894 KB)
[v2]
Tue, 22 Sep 2026 03:24:28 UTC (883 KB)
[v3]
Sat, 24 Oct 2026 21:04:57 UTC (884 KB)
来源：arXiv cs.AI · arxiv.org