You must log in or register to comment.
The article doesn’t really go into detail, but having a “synthetic dataset to negate” the personal data, sound to me like they still keep the personal data around (just somewhat obfuscated), but found a “clever” way to scrub it from the machine learning model’s output…