You have definitely had multiple scenarios where you finish having a conversation with someone, and opening your phone, you see an ad that relates to the conversation. You can even claim that you saw an ad for a product that you only thought about. This seemingly ‘supernatural’ ability of our devices to predict our desires has spawned countless social media posts and conspiracy theories. These concerns aren’t unfounded – Big Tech’s extensive history of privacy violations has created a foundation of mistrust. So when you receive ads eerily similar to your offline conversations or activities, is your phone spying on you? The answer is more nuanced than a simple ‘yes’ or ‘no’. the answer is both ‘Yes’ and ‘No’.
Voice Assistants
You’re not entirely wrong – your phone does have the capability to listen and it does listen, just not in the way you might think. Modern smartphones (and nearly all smart devices) come equipped with voice assistants that let users control their devices through simple voice commands. At their core, voice assistants operate on a principle that’s both simple and sophisticated: the ‘wake word’ system. Rather than recording everything you say, these assistants use a small, specialised processor to listen for specific trigger phrases: “Hey Siri,” “OK Google,” or “Alexa.”
However, there is a crucial distinction between active listening (waiting for a wake word) and passive surveillance (constantly recording conversations). For the voice assistants to work, your device isn’t recording everything – instead, it’s continuously processing a small audio buffer, listening specifically for pattern matches to wake words like “Hey Siri” or “OK Google.” Think of the audio buffer as a small loop of tape that’s constantly being recorded over. When you say the wake word, only then does the system start actually saving and processing your voice command. As such, they do not transmit audio while passively listening. Most devices provide visual feedback during active listening – an animation, a light, or an on-screen indicator, depending on your device.
These voice commands are typically stored in your account history, which you can review or delete at any time. The ideal state of any product is a feedback loop; the first iteration of the voice assistant is shipped, and voice data from customers are collected and used to improve the service. As the service and offerings improve, more people opt-in and more data is collected, further refining it. Usually, this data is anonymised and these companies have explicitly stated they do not use voice data for targeted advertising. Users typically have the option to opt out of contributing their data to these improvement programs.
Apple recently agreed to pay $95 million to settle a lawsuit filed in 2019 citing privacy concerns including accidental activation of Siri leading to private conversations being recorded.
The controversy centred around third-party grading – while Apple disclosed that they collected recordings, they hadn’t explicitly mentioned that humans would review these recordings.
Despite widespread concerns, there’s no concrete evidence that major tech companies like Apple, Google, and Meta use voice data for advertising profiles. So if those eerily target ads do not come from spying on your calls and other voice data, how do they come about?
Big Tech and Digital Tracking
When Instagram seems to read your mind, showing ads for a tent right after you discussed travelling, it’s not the voice listening – it’s surveillance capitalism at work. Let’s examine Meta’s Pixel to understand how companies track and target you without ever needing to listen to your conversations.
Meta’s foray into extensive user tracking began with a fundamental question: how could Facebook gain insights into user behavior beyond its platform? This led to the creation of a tool called Facebook Pixel now known as Meta Pixel. Pixel is advertised as a piece of code added to your website that lets you analyse traffic to your site to optimize traffic. What Pixel actually does is watch what you do on websites that have this piece of code embedded and report that information back to Meta. This data covers everything; how long you linger on a certain webpage, what you add to your cart, what you buy and so on. Crucially, location data plays a significant role in this tracking, as it reveals much about an individual’s lifestyle and preferences. By extracting location data from mobile devices, check-ins, IP addresses, and other sources, Pixel constructs comprehensive profiles used to deliver targeted advertisements.
Each time a user accesses Facebook, using a specific browser, machine, and IP address, this generates a signature that Facebook matches to that user. If the same user uses several devices and several browsers (as we all do), all of these combinations can be associated with one, single user. The platform further analyzes social connections, understanding that friends and people around you (based on location) often share similar interests and purchasing patterns. When several people in your network engage with related content, the algorithm predicts you might be interested too.
For example, let’s say Jane is travelling internationally to see her fiancee. Upon getting to the airport, she realizes that she forgot her sweater and looks up options. If the site she visited has Pixel, chances are, her fiancee will receive ads for sweaters based on location correlation. If the air conditioner in your apartment building goes out and occupants start looking up air conditioning, you’re likely to receive ads about air conditioners.
This extensive tracking is not exclusive to Meta; other tech giants like Google employ similar strategies by monitoring browsing history and location data. Big Tech companies are not only interested in online behaviour but also seek insights into users’ offline activities. They achieve this by collaborating with data brokers and advertising networks that sell additional information, allowing them to create detailed consumer profiles. Modern advertising techniques are so advanced that they can anticipate significant life changes before individuals publicly disclose them. For instance, Target was able to predict pregnancy and how far along shoppers were based on shopping patterns as early as 2010 and used this information to tailor advertisements accordingly. A lot of what is described as “listening to conversations” can be explained by the correlation-to-causation pipeline.
The ads feel psychic, but they’re actually the result of analysing thousands of data points about your behaviour. This comprehensive tracking system explains why ads feel so personal without any need for audio surveillance. Meta and other tech companies have built something far more powerful than an eavesdropping system – they’ve created predictive algorithms that can anticipate your needs based on your digital footprint and social connections.
This is all to say your activities online and offline represent valuable data that advertising can harness to capitalize on your purchasing power. So why don’t companies just record your conversations? The scale of such an operation would be staggering. To serve targeted ads based on voice recordings, they would need to store approximately thousands of petabytes of new data daily. To put this in perspective, this would be like trying to store the entire printed collection of the Library of Congress hundreds of times every single day. The infrastructure required to store this magnitude of data would be astronomical, both in terms of physical space and cost. Taking into account the manpower, legal and financial implications of continued listening, it’s clear why these organisations would employ sophisticated data analytics that translates your digital body language into ads rather than attempting to implement continuous audio surveillance.
So now, answer the question for yourself: is your phone listening?