Tom Philip

Privacy is a big deal in voice, but it’s sometimes frustratingly hard to get clear answers about what goes on with our data.

I grew up in what could reasonably be described as a technophobic house: My parents did not own or even entertain the idea of microwaves, believing the fast cooking times and the scary word “radiation” would eventually poison us all. They would turn the wifi router completely off at night, unplugging it from the wall, not wanting to be blasted with “beams” while they slept. 

Safe to say, they do not own a smart speaker, nor do I think I could even bribe them to live with one for a week.

There is, understandably, a lot of concern over “always-on” microphones as it relates to our privacy in our own homes, but there’s also plenty of exaggeration and misinformation. Here are the most common types of smart speakers you’ll find in the wild, and how they handle your information.

Amazon Echo

By far the most popular smart speakers in the US, the Echo and its built-in assistant, Alexa, are pretty much ubiquitous with voice tech at this point. Amazon’s made strides with its commitment to privacy: the 4th Generation Echo has a dedicated “microphone off” button that will disable you from waking Alexa in any way until you press the button again. 

The Echo, like any decent smart speaker, is reliant on a “wake word” to know when to start listening. So unless someone in your house is named Alexa, there’s a very, very low chance you’ll ever be recorded saying something you’d like off the record.

As for those records, Amazon *does* store plenty of data from your speaker, but only very small portions of it would or could ever be reviewed by another actual human. Amazon employs a batch of contractors to listen to a small portion of recordings to help refine Alexa’s responses to a request from someone with a thick accent, say, or a request to rent a foreign film with a difficult-to-pronounce name. The overwhelming majority of voice recordings ever used by Amazon are to help the user experience. But, yes, the echo *can* pick up sound clips, and sometimes that data *can* be sent to the authorities in the most extreme cases, like when the police think the Echo might have recorded a murder. The good news is you can also opt out of having any of your recordings used by a real human in the Echo’s privacy settings.


Apple’s embarrassing leak two years ago, in which a contractor listening to recordings went public about the kinds of conversations (or, uh, more *intimate* encounters) employees sometimes listened in on, might just have been a blessing in disguise. 

It rightly caused concern across the cultural spectrum, and both Apple and Amazon shored up their commitment to privacy in its wake. While it looks like both companies have continued to use human workers to analyze a tiny percentage of recordings, they’ve made it much clearer to the average user about how and why data is collected, and guarantee total anonymity, insisting no other identifying information is attached to its recordings.

There’s also the issue of Siri, which lives inside the phones of a huge portion of the population. There have been reports here and there of people’s iPhones serving up ads for things they might have mentioned in passing while their phone was in their pocket but there’s no hard evidence to suggest that’s the case. I was once walking with a friend and pointed out a chicken coop I’d helped paint one summer (very Norman Rockwell). 20 minutes later she got an ad for a chicken coop on her Instagram feed. Was Siri listening? Did another app triangulate her location during the walk and notice she’d passed a chicken coop? Was it all a big coincidence? It can be annoying to trawl through all the privacy jargon each of these companies attaches to its products, especially since they’re ever-changing, but you’ll be glad you did the research if it helps you become more comfortable interacting with new tech. 


About six months ago Google updated its privacy policy so that only people who *opted in* would ever have the small chance of their voice recordings being sent to a human to review. Again, Google maintains no other information about you is included with these recordings.

The bottom line

Honestly, in researching this piece, I was expecting to discover far more horror stories about accidental recordings, embarrassing mishaps, and even scary surveillance, but we’re now more than a decade into consumer voice tech, and while there have been some hiccups, it’s nothing you wouldn’t expect from any variety of bleeding edge new technology. 

As with most things, you should do your research before immersing yourself in any kind of “smart” appliance or application. Voice AI has done wonders for accessibility for disabled people, for example, but others might not see the use in a glowing orb that can tell you what the weather will be like tomorrow. 

Simply, it comes down to a case of what you want. Of course there are downsides to voice tech, and the big companies don’t have the *best* track record with being utterly truthful, but it’s safe to say plonking an Echo in the middle of your living room won’t be the first domino that falls in a Skynet-style robot uprising. Be safe, be smart, and put your gadgets to good use.