Who is already labelling.
Sites listed through the plugin opt in from their own dashboard and pass a token handshake, so those listings are self-attested and verified, not scraped. Our own two sites are marked as such: they are static sites rather than WordPress, so they cannot run the plugin or the handshake, and their counts are the images that actually carry IPTC provenance, counted from the files.
Listings come from the registry API on the live directory. Any site running the free plugin can opt in from its dashboard; the edge verifies control of the domain before it appears.