Meta grants itself right to scrape content from business customers for model training
(www.theregister.com)
Except there's no guarantee they'll be coming from one of their own IP addresses. There's nothing to stop them from using a residential botnet.