> because it would be doing so without having access to the source

I find it unlikely that photopea was never scraped for AI training considering they are looking so hard for new material they started buying up and scanning old books.