Artificial Intelligence and Misinformation in Art: Can Vision Language Models Judge the Hand or the Machine behind the Canvas?

This repository contains a collection of AI-generated images created from paintings in the WikiArt dataset

Repository Structure

• All_gpt4.1-mini_prompt.xlsx: Prompts generated by GPT-4.1-mini for each original artwork, used as input for text-to-image models.
• images/:Original artworks from the WikiArt dataset (excluding works by “Unknown Artist”), along with their corresponding AI-generated paintings based on GPT-4.1-mini descriptions.

Dataset Preprocessing

The WikiArt dataset is sorted by artist name using the following code:

from datasets import load_dataset

ds = load_dataset("huggan/wikiart")
ds["train"] = ds["train"].sort("artist", reverse=True)

Citation and paper

https://dl.acm.org/doi/10.1145/3807498 

nez, Gonzalo and Reviriego, Pedro and Merino-G’omez, Elena and Moral-Andr’es, Fernando},
title = Artificial Intelligence and Misinformation in Art: Can Vision Language Models Judge the Hand or the Machine behind the Canvas?,
year = 2026,
issue_date = June 2026,
publisher = Association for Computing Machinery,
address = New York, NY, USA,
volume = 19,
number = 2,
issn = 1556-4673,
url = https://doi.org/10.1145/3807498,
doi = 10.1145/3807498,
journal = J. Comput. Cult. Herit.,
month = jun,
articleno = 33,
numpages = 25,
}