Documents released in copyright litigation against OpenAI and Microsoft showed internal scientist pushback over web-scraping fair use claims.