Does Your Phone Listen In On Your Conversations?

Have you been targeted by an ad for something you’ve only spoken about out-loud? Your phone could be listening in to your conversations.

Everyone who owns an iPhone seems to have at least one story of where they claim to have been targeted by ads based purely on what they had been saying out loud. A conversation with friends in the pub about a particular make of shoe might end with that exact shoe popping up while you’re on Instagram or Facebook the very next day. Even when you could have sworn you’d never actually typed the product name into your device.

One iPhone user tells The Versed about an afternoon in the park with a friend who introduces them to a book publisher they’ve never heard of. “The next day, I saw an advert on Facebook for them. I’d no idea they existed until my friend told me about it the day before so there’s no way I could have typed it into google. It was pretty unsettling to think about the possibility of being listened to and targeted with products based on that.”

Stories like this are not uncommon but are they based in reality? Or is data-based marketing now accurate enough that it targets you with things you’re likely to talk about?

Photo by Mizter


The technology needed for this type of audio-tailored marketing is very much in place already. Features like ‘Siri’ or ‘Google Assistant’ are triggered with ‘hey Siri’ or ‘okay Google’ without the user needing to touching their phone.

These voice recognition services are highly personalised – they’re trained to only listen to your voice – and can be used to send texts, make calls, play certain types of music, and countless other tasks.

It would be very easy for this technology to ‘listen’ to a conversation of yours, make a record of brand names mentioned, like ‘Adidas’ for example, and also subject matter, ‘Sport’ and then use that data to target you with very personalised adverts. Phones are definitely able to do this, but whether or not they actually do is in dispute.

Facebook and Apple have continuously denied accusations of ‘listening in’ to user conversations and targeting ads based on audio collected in this way. But one tech security expert argues that they could be.

Dr. Peter Henway told Vice that from time to time “snippets of audio” go back to apps like Facebook “but there’s no official understanding what the triggers for that are” he says, a ‘trigger’ being a word or phrase that ‘triggers’ the phone to start listening to you.

Photo by Craig Whitehead

Phantom Advert

The general consensus among tech security experts is that while these companies certainly have the capabilities to listen in on conversations, and potentially even the legal right to do so, they probably aren’t doing it. With regards to advertising, they don’t really need to. The data available to these companies is more than enough to create an incredibly detailed picture of an individual without the need for audio information too.

For example, location settings are something most smartphone users will have enabled, and location data can be incredibly revealing. Taking the book publisher case mentioned above as an example,

A user doesn’t need to say the name of the publisher out loud for them to see ads about it, Facebook knows that the user has just spent 4 hours in the park with a friend who’s bought from that publisher many times, and therefore, the user is also likely to be interested, and so they start seeing ads for it.

In short, sufficient data is available to these companies already in order to push highly accurate targeted advertising to its users.

Following repeated claims, Google recently stated that they “do not use ambient sound from any device to target ads.” Facebook has also said in the past that they “show ads based on people’s interests and other profile information – not what you’re talking out loud about.”

Regardless of what’s really happening, what’s for sure is that users will continue to report disconcertingly precise adverts popping up on their timelines, and social media companies will probably continue to deny using audio to reach those levels of precision.

Start the discussion

to comment