Oh I'll agree. LLM's are an exceptionally efficient tool for gathering, collecting, and organizing that kind of data.

Most of it wouldn't be anywhere near as fucky if there were proper privacy protections for citizens, both from governments and corporations. Or if not for the attacks on VPN's and the attempts to push digital ID's.

Although even then, I do admit that AI could still be used to identify an individual (writing style analysis, any scraps of identifying info leaked from your connection) unless you maintain every security precaution imaginable every single day you use the Internet.