They may not be eavesdropping with the microphone but they could be with an accelerometer[0].

  In particular, we show that by exploiting the affected accelerometer readings and carefully selecting feature sets along with off-the-shelf machine learning techniques, Spearphone can perform gender classification (accuracy over 90%) and speaker identification (accuracy over 80%) for the audio/video playback on the smartphone for our recorded dataset.

Google was capable of warning people of an earthquake before it even happened[1].

  Google has also turned individual phones into miniature earthquake sensors. All smartphones have accelerometers that can pick up signals of an earthquake. If triggered, the phone sends the message to a detection server, along with rough location data, like the city a device is in. The server then pieces together where the earthquake is happening from data collected on multiple phones and beams out the relevant alerts.

[0] - https://arxiv.org/abs/1907.05972

[1] - https://arstechnica.com/science/2022/10/how-google-alerted-c...