Effectiveness of Image Protection Software Against Image Generation Tool Training

Abstract

Unauthorised use of artworks in training image generation models poses a growing challenge for online copyright protection. Contemporary artists’ work is used without their consent both by artificial intelligence (AI) companies and Internet users, creating an urgent need for effective protective measures. The article examines and compares software solutions designed to safeguard artworks from such unauthorised use, examining both technical effectiveness and the output’s visual quality. We evaluate different image protection tools and their change intensity levels by applying them to illustrations created by one of the authors and then training generative models on both protected and unprotected versions. The resulting images are evaluated by a voluntary survey of 71 respondents including artists and artwork viewers (non-artists). The discussion and conclusions assess image protection software based on research findings, provide recommendations for artists, outline future research opportunities, and demonstrate that participation increased respondents’ awareness of the importance of protecting artworks.