I've noticed that after discussing a topic with family, friends, or coworkers, I sometimes see related videos and recommendations on YouTube, TikTok, and other apps. The unusual part is that my own phone was not always nearby. For example, at work we discussed Ronaldo Nazário's injury and surgery while my phone was locked in my car. Later, recommendations about him and the injury appeared on my phone. Could another person's phone have captured the conversation and somehow connected it to my advertising or recommendation profile? Or are shared locations, search history, social connections, recommendation algorithms, coincidence, and the frequency illusion more likely explanations?
4 Answers
Advertising systems can infer that devices are associated through shared locations, recurring proximity, Wi-Fi networks, contacts, or other behavioral patterns. If one person searches for a subject and another device is repeatedly linked to them, the system may show related content to both. That still doesn’t require recognizing your voice or knowing exactly who was part of the conversation; it can be a statistical guess, and sometimes the recommendation is simply a coincidence.
Claims that every microphone constantly sends all nearby conversations to advertisers are much stronger than the available evidence supports. Voice assistants may process audio for wake words, and malicious apps could misuse microphone permissions, but ordinary recommendation systems generally have plenty of other data to work with. Checking microphone permissions, disabling ad personalization, reviewing location history, and comparing recommendations across devices can help test whether the pattern is real.
The timing can feel extremely convincing, but recommendation feeds are also full of popular or currently trending subjects. If someone had already searched for Ronaldo, the platforms might have been promoting that content broadly anyway. Without controlled tests—such as discussing uncommon subjects while preventing nearby devices from accessing them—it’s hard to distinguish targeted collection from algorithmic prediction and selective attention.
It’s technically possible for phones to record audio, but there’s no good evidence that major platforms routinely identify nearby strangers by voice and attach conversations to their individual advertising profiles. That would be difficult to do reliably and would create obvious privacy and battery-use problems. A more plausible explanation is that someone nearby searched for the topic, several people were exposed to related content, or the recommendation system made a broad prediction based on location, activity, and existing interests. Once the topic is on your mind, the frequency illusion can also make related recommendations stand out more.
I’m not convinced that fully explains this particular example. There were many employees nearby, and my phone was in my car. I’m wondering how a system could identify me specifically rather than simply recommend the topic to everyone in the area.