Abstract: Artificial intelligence has become one of the most widely discussed topics of our time. These discussions have shown that while rapid technological advancement brings positive outcomes, it also raises some ethical concerns. These concerns are especially emphasized on the topic of generative artificial intelligence. Using personal photos of people on the training datasets of AI applications, or manipulating these kinds of photos has led to misinformation forming on a societal scale. The topic of AI applications using original artwork for their datasets has also been discussed widely among artists. Artists share a common view that the situation is not ethical, and they feel that their efforts are being taken advantage of in this way. In line with these concerns, many approaches have been proposed to minimize the negative effects of artificial intelligence. One of the methods includes poisoning the images that might be used in a dataset, which concludes in the artificial intelligence being unable to generate similar images. In the application discussed in this article, the aim is to expand this method and ensure the protection of personal data and intellectual property rights. In line with this goal, we propose an application that protects both personal images and original artworks from being manipulated and used in datasets by artificial intelligence programs.

Keywords: Generative artificial intelligence, privacy, dataset poisoning, ethics, Generative Adversarial Network

Pages: 252-257

WSEAS Transactions on