The issue of AI scraping copyrighted works to train language models is heating up, with Dow Jones and CNN starting to talk about charging companies like OpenAI when their chatbots lap up their work. Is this infringement, with the resulting AI a derivative work that requires licensing? Or is it instead a fair use that requires no payment, because arguably the copying is “de minimis” and the AI generates something transformational (which is the relevant copyright test)?
Either way, artists, creators and publishers justifiably feel threatened from both an economic and a moral standpoint. The economic threat is clear: Their creative livelihoods are in the crosshairs.