This mass surveillance algorithmic detector tool, Semantic Forensics (SemaFor), has in fact been developed and is practically good to go. It claims that this project will be used to detect AI deepfakes such as, for example, those used for fake news disinformation.

Considered by many to be untrustworthy, might DARPA be a major part of the disinformation problem…??

Bear in mind that those overseeing this want your acquiescence, wanting you to accept that they will be scanning every related aspect, including your personal data, to distinguish what's real from deep fakery. After all, they want you to think that you don't know any better…

They claim that they will collect the scanned data to look out for disinformation, particularly, falsehoods for example, in the form international "large-scale disinformation attacks."

Take a look at this DARPA video promoting this mass surveillance project: