Previously, Amazon’s smart speakers were found to be quietly recorded even if the user did not interact with them, and the content was returned to Amazon for analysis after recording. At the time, Amazon said that listening to user sessions was mainly used to improve speech recognition, but Amazon did not inform in advance of the existence of such quiet recordings. Simply put, when the user is not interacting with the smart speaker, various conversations and other sounds in the home environment are recorded and uploaded to the server.
Recently, anonymous users have cooperated with the Belgian broadcaster VRT News to show evidence of Google’s stealing of user privacy: Google’s smart speakers are also quietly recording user voices. Google Speaker is also the Google Home series, and its built-in Google Assistant responds to various voice commands after the user says OK Google.
However, the fact is that even if the user does not say OK Google Google speakers are recording, these sound clips may be edited and then uploaded to the Google server. The same Google did not inform the user in advance that the audio might be quietly recorded, and then all the users were monitored by Google speakers around the clock.
In response to the above news, Google has issued a statement saying that the company will actually use Google speakers to record the sound of the home environment without using commands. These sound clips are automatically uploaded to Google servers to improve Google’s speech recognition software to more accurately identify content and commands.
At the same time, Google also said that the sound clips have been anonymized to remove content that may contain user information, and of course, some privacy sounds will not be uploaded. But Google can also use its network to identify enough information about users, such as the user’s real name and home address or even cell phone numbers and other private information.