The app continues to collect microphone audio even if the microphone is muted during a video conference

When you are temporarily away from your desk to get things done during a video conference or voice chat, you may want to mute the microphone to prevent extra voice from reaching the other party. However, 'If you mute the microphone during a video conference, the app itself collects audio data, even if the other party doesn't receive the audio,' said Kassem Fawaz, assistant professor of electrical and computer engineering at the University of Wisconsin. Announced.

Are You Really Muted ?: A Privacy Analysis of Mute Buttons in Video Conferencing Apps
(PDF file) https://wiscprivacy.com/papers/vca_mute.pdf

You're muted — or are you? Videoconferencing apps may listen even when mic is off
https://news.wisc.edu/youre-muted-or-are-you-videoconferencing-apps-may-listen-even-when-mic-is-off/

The reason for the research was that when Assistant Professor Fawaz's younger brother muted the microphone with the function of the app in the video conference, he noticed that the lamp of the microphone was still lit, and his microphone was accessed even while muted. He said he consulted with Assistant Professor Fawaz because he was worried about it.

A research team led by Assistant Professor Fawaz conducted an awareness survey of 223 video conference app users regarding their understanding of the mute function and the handling of audio data. As a result, we found that most users think that 'the app should not be able to collect audio data while the microphone is set to mute.'

In addition, the research team uses a run-time hex analysis tool to track the audio data of major video conferencing apps released on Windows, macOS, iOS and Android, and the audio app while the app is muted. I investigated how to move from to the computer's audio driver to the network.

Investigation revealed that all the applications tested were collecting audio data even when muted. We also found that some apps were sending the collected data to the server.

So the research team used the voice data collected from the app to see if they could infer what the muted user was doing. When we analyzed the collected data using a machine learning algorithm that learned daily life sounds such as cooking, eating, music playback, typing, and cleaning from YouTube movies, we identified activities that were muted with an average accuracy of 82%. It seems that he was able to do it.

Assistant Professor Fawaz said, 'If you turn off the camera or hold your hand over it, you won't be able to see anything, but with a microphone, that's not the case.' He argued that the solution would be to develop a switch that could disable the microphone and software that could easily disable the microphone without accessing the OS settings.