Let's be fair about the tracking: basically every modern car sold in the US and Europe is being tracked through built-in SIMs. And even older cars are mostly tracked through mobile phones, and in the US through Flock and similar surveillance networks. Ad-tech companies have built up detailed profiles, frequently including detailed location data. And we have no reason to assume our governments aren't actively tracking all these sources.
With today's AI models, it'd be feasible to analyze massive amounts of video/audio data gathered by the microphones and cameras all around us - as long as you have enough GPU data centers.