This isn't a new revelation, ChatGPTs terms of service forbids using it's output to train competing models. Users can choose to publish their chat logs directly on chatgpt.com where they probably will get picked up by competing scrapers unless they are blocked.
No, the hypocrisy of OpenAI handwaving their own scraping as "fair use" while classifying scraping their own services as "abusive behaviour" is not lost on me.
No, the hypocrisy of OpenAI handwaving their own scraping as "fair use" while classifying scraping their own services as "abusive behaviour" is not lost on me.