What prevents a malicious actor from scraping the Flickr new photos feed (for example) of someone else, and registering & selling that other person's copyrighte

What prevents a malicious actor from scraping the Flickr new photos feed (for example) of someone else, and registering & selling that other person's copyrighted work? Or a malicious actor posting their own photo to Flickr, then registering it on ascribe, selling it through ascribe, and then repudiating the transaction, claiming they weren't the ascribe user? Unless ascribe.io is the first place something is ever published, or they have some way to verify initial ownership (not just possession), I don't see how this works.