My slightly vague recollection was that they were basically feeding "enterprise customers" a load of information including stuff that could be used for union busting, monitoring protests etc. Their enterprise plan has
Feedly AI Advanced Skills: Market intelligence Threat intelligence Biopharma research Competitive intelligence
as features. So yeah, creepy as fuck. And they said at the time that this was all done using "AI".
Of all the companies, Google always seemed the most likely, both to want to and to be successful. They’ve tried before, sometimes in small ways, sometimes in larger more obvious ways (AMP, the implementation of content filtering in Chrome etc.).
They’re the world’s largest advertising and data harvesting company. It’s their business. Of course they want to lock the internet down to serve their goals of learning as much about you as possible and using that data to shove ads in your face.
Whenever using any Google/Alphabet product you have to ask yourself, “am I ok with this thing I’m about to use being built by the world’s largest advertising company?”. The answer should be “no” more than it is “yes”, particularly for things that have access to lots of your data, like web browsers, phones, home speakers etc.