I’m not trying to be obtuse - I understand the risks here - and I don’t want to get into an argument about semantics - but

Populating a variable based on the current user so as to segment them in analytics or advertising - regardless of whether it’s that the user is pregnant or they said their favorite color is blue - is different than “sharing” structured pregnancy data or favorite color data that a 3rd party can inspect.

If this sort of wide-scale actual sharing of data exists, then I’d expect to see job postings for the role of coordinating data exchange, schemas for pregnancy data in XML format, API documentation for transmitting ‘sensitive’ data to 3rd parties, leaks containing this sort of sensitive data in unexpected places, etc.

That said, the title of the article says “profiling” and not “sharing”, so maybe Facebook and Google can derive that particular theoretically opaque segmentation values from particular apps represent pregnancy status.

Or maybe an opaque, trained model outputs that for certain users with certain segmentation values in certain apps, those users are more likely to click on ads relating to pregnancy.

Edit: The WSJ article notes that Facebook has acknowledged that the latter is possible via a filed patent. FB say it doesn’t use that technique.