Video chat apps could still be listening to you even when you are on mute, researchers say

Man gets rejected from job because of tweet joking about setting Whole Foods ‘ablaze' (Getty Images/iStockphoto)

Video chat apps could still be listening to your voice even when you are on mute, new research suggests.

Examination of a variety of popular meeting apps found that they were still collecting audio even when a user thought they had switched off their sound.

Some of that information is being sent up to the cloud and so could be listened to by those companies, they say.

And it is possible to work out even what people are doing using that sound, the researchers warn.

Researchers at the University of Wisconsin–Madison began the research after being alerted to the fact that video chat apps appeared to be accessing the microphone even when they claimed not to be. Researcher Kassem Fawaz was told by his brother that his computer was showing a light indicating that the microphone was being used even when the mute button was pressed.

Some operating systems offer a warning when a given app is accessing audio. Apple’s iOS and MacOS, for instance, shows an orange light in the corner of the screen if sound is being recorded by an app.

That led Professor Fawaz and a graduate student, Yucheng Yang, to examine popular video chat apps on a variety of platforms to understand whether they stopped listening to their users.

“It turns out, in the vast majority of cases, when you mute yourself, these apps do not give up access to the microphone,” said Professor Fawaz in a statement. “And that’s a problem. When you’re muted, people don’t expect these apps to collect data.”

The researchers analysed the code powering a variety of apps to examine how the audio was sent around the computer when they were being used. They found that all of the apps in the test were gathering raw audio when they were on mute – and that one of them was sending that information over the internet at the same rate, whether the user was muted or not.

They even found that it was possible to use machine learning to gather that audio and quickly assess what people might be doing while they were in the app. They found it was possible to identify whether the person was cooking, eating, or typing, for instance, with 82 per cent accuracy.

Even without any indication that the data is being used by companies in that way, it is an important warning that the apps could be collecting private information without a user being able to stop them, researchers said.

“With a camera, you can turn it off or even put your hand over it, and no matter what you do, no one can see you,” said Professor Fawaz. “I don’t think that exists for microphones.”