These tools seem to be using others’ content for more than just training. There is growing anecdotal evidence that the content the AI creates contains recognizable segments/samples of other creators’ work. As in, chatbots recite entire sentences and paragraphs of other work in their answers, and image generators replicate parts of known images from other artists in their output. Not sure that qualifies as fair use, especially if OpenAI seeks to profit off the content its technology generates.