Artists Want to Prove Their Work Was Stolen by AI. A New Study Says That’s Impossible

Artists are facing a significant hurdle in their efforts to prove that AI has stolen their work. A new study has cast doubt on the ability to trace generated images back to their original training data. This development has implications for artists who believe their creations have been used to train AI models without consent or compensation. The study, published by MIT News, indicates that the links between AI-generated art and the original creators are becoming increasingly difficult to establish.

This is due in part to the growing and evolving datasets used to train AI models. Fast Company and Quantum Zeitgeist have both covered the study, noting that the debate over whether generative AI actually copies artists is far from settled. The complexity of AI training data makes it challenging to definitively prove that specific artwork was used in the process. Artists seeking to protect their intellectual property may find themselves in a precarious position.

According to Straight Arrow, MIT's research suggests that proving the use of specific artwork in AI training may be impossible. This leaves artists with limited recourse if they suspect their work has been appropriated by AI systems.