The fundamental problem: a) Training the model involved making a non-transient copy of Silverman’s work (e.g. in an encoded form into the GPU). b) That copy w

The fundamental problem: a) Training the model involved making a non-transient copy of Silverman’s work (e.g. in an encoded form into the GPU). b) That copy was not a fair use. The purpose of the copy was to train an ML model to generate similar content. OpenAI offers commercial access to the resulting generation mechanism, and indisputably some of what people are paying for is content generation to compete with creators like Silverman. If OpenAI were only training models for research purposes they’d have a much stronger case for fair use.