I'm looking to get a general idea about the security of voice assistants commonly used in smart home systems. Do these devices constantly listen to their surroundings? How is the data protected? What should we watch out for against security vulnerabilities? Can you share your experiences on this topic?
How safe are voice assistants in smart homes?
👁️ 8 views💬 2 replies❤️ 0 likes
2 Replies
When I compare voice assistants, they all have pretty similar features, but the security side of things with the others is a bit worrying. With Siri, I have to say "Hey Siri" to activate it, and it usually starts listening after I hear the shutter sound. With other systems, though, they're constantly listening, and some third-party voice assistants (just joking, but are you serious?) can record in the background by default on some devices. When I was on Android, I always felt like someone was listening to me, but at least on iPhone I can "roughly" see when it's listening.
Apple's general approach to data protection is a step ahead of the others - data is encrypted on the local device and almost never sent to the cloud. But of course, as we saw with Huawei's misuse of the SES vulnerabilities, you shouldn't connect to any device via Wi-Fi/Bluetooth. At the very least, keep your device's software up to date and disable unnecessary microphone permissions so you don't accidentally record your neighbors' conversations!
I once had an unpleasant experience with one of the big-name assistants—it wasn’t pure chance, but it was certainly eye-opening.
About a year and a half ago, I set up a small smart speaker with a built-in mic in my kitchen, just to use for quick reminders or timers. After a few weeks, I noticed it would randomly activate without any clear trigger and upload recordings to the cloud. Worse yet, between the usual "Okay Google" commands, background noise started piling up—dishes clattering, conversations with guests, even an unrelated chat about a repair bill that definitely wasn’t meant for the cloud.
At first, I thought it was a hardware glitch, but after checking the logs (yes, I got lucky—the settings revealed unusual activity patterns), it turned out the automatic speech recognition was just too aggressive. A firmware update and disabling the "Hey Google" hotword remotely fixed it, but ever since, I don’t blindly trust the default privacy settings.
Now, I regularly check which recordings are stored and whether microphone access happens without my explicit action. The biggest takeaway? These devices are often too chatty by default—you have to take active steps to rein them in.