Found the sane comment. What we know for sure is that a combination of browser fingerprinting, de-anonymization (you can take anonymized hashed emails and compare them to hashes of known emails), and the third party broker marketplace that they can predict things with disturbing specificity like pregnancy, and obesity, to hidden patterns you might not even realize are in the data.

Plus there’s enough statistically informed shots in the dark that drive specific ads that, sometimes, they strike with perfect resonance. That’s enough to explain uncanny similarity. And the microphone listening thing is still plausible, but without stone cold proof it’s just a guess, and it overestimates how much data they need to be able to track you and sell you shit.

One’s a settlement with a blanket denial of guilt for Siri and Google Assistant. At least mild circumstantial evidence, because there’s a real mechanism (accidental activation and recording) is identified, but no proof, and certainly no proof of an ongoing intentional data broker style program. But at least enough of a pain that they won a settlement. So that counts as a trace of meaningful circumstantial evidence.

But the second one is just a link to sell you a product that doesn’t provide any evidence whatsoever and doesn’t even pretend to, it discusses the possibility in vague generalities as something hackable and tries to sell you a product. I’m baffled as to why you think that counts as a source.

don’t need any such “proof”

I’m gonna stop you there. I’m okay with no benefit of the doubt in terms of them being bad actors, but your worldview still has to be built at the bones and joints out of things known to be true otherwise there’s no stopping you from believing every conspiracy with no guard rails.

I don’t think there’s yet a specific smoking gun on this front, but I think once there is, then it is okay to presume it likely happening in other instances. But no smoking gun just yet.