Not with my name! Inferring artists' names of input strings employed by Diffusion Models

TL;DR

This paper investigates whether diffusion models generate images based on artists' works by inferring the likelihood that an artist's name was part of the input prompt, using a Siamese Neural Network on images from DALL-E 2.

Contribution

It introduces a novel method employing a Siamese Neural Network to estimate the probability of an artist's name being in the input prompt of generated images.

Findings

The approach effectively predicts the likelihood of an artist's name in input prompts.

Experimental results show the method as a promising starting point for further research.

The dataset and code are publicly available for reproducibility.

Abstract

Diffusion Models (DM) are highly effective at generating realistic, high-quality images. However, these models lack creativity and merely compose outputs based on their training data, guided by a textual input provided at creation time. Is it acceptable to generate images reminiscent of an artist, employing his name as input? This imply that if the DM is able to replicate an artist's work then it was trained on some or all of his artworks thus violating copyright. In this paper, a preliminary study to infer the probability of use of an artist's name in the input string of a generated image is presented. To this aim we focused only on images generated by the famous DALL-E 2 and collected images (both original and generated) of five renowned artists. Finally, a dedicated Siamese Neural Network was employed to have a first kind of probability. Experimental results demonstrate that our approach is an optimal starting point and can be employed as a prior for predicting a complete input string of an investigated image. Dataset and code are available at: https://github.com/ictlab-unict/not-with-my-name .