FYI, we're still not sure whether the scraped AI training datasets involve copyright infringement.