A worthwhile essay by Trystan S. Goetze on the ways generative AI constitutes theft:
…built on the appropriation of vast troves of data obtained without consent. Moreover, machine learning models that are not employed in generative systems—such as, perhaps, certain large natural language understanding models— may also involve theft. New approaches to data collection and processing are needed for generative AI development to be morally permissible. Until then, these impressive new technologies do not stand on the shoulders of giants; rather, they parasitize their innards.
AI Art is Theft: Labour, Extraction, and Exploitation—Or, On the Dangers of Stochastic Pollocks 🎨