Google Home Products Listening In
Google Home products usually spend most of their time waiting for activation phrases like “Hey Google” or “Ok Google” before offering their assistance. However, artificial intelligence-based assistants aren’t always that “intelligent” and regularly – and sometimes hilariously – miss the mark. That’s why tech companies still employ humans to analyze user data and audio recordings. Humans can, for example, compare what the AI-based assistant thought it heard with what the user asked. They can also verify if the answer given actually made sense. Next, developers use this type of information to improve their products and/or develop new features. In theory, all user data is anonymized. However, the content may still reveal a great deal about the user, such as the user’s interesst and/or their location, for example.
Opt-in “Accidentally” Switched On
However, during a recent Google software update some of these privacy features were being ignored. On 5 August, 2020, Google updated the settings for voice and audio recordings. In an e-mail to users, Google confirmed that consumers have a choice: “To keep you in control of your audio recording setting, we’ve turned it off for you until you are able to review the updated information. Visit your Google Account to review and enable the audio recording settings if you choose.” Similarly, in April 2020, Google made it possible for users to configure the sensitivity of Google’s Assistant. This is to ensure it does not accidentally activate, and starts recording, when it thinks it heard a wake-up word. Again, this feature is intended to give consumers more control over Google’s privacy settings. It allows them to determine when and which data is being recorded and saved to their Google Account. Google claims that the opt-in feature had accidentally been switched on during the software update. The company confirmed that this mistake had since been corrected. Yet very recently, another concerning fact came to light. Apparently, some Google Home devices are being activated without any activation command being spoken and start recording random sounds. Even more bizarrely, and contrary to Google’s earlier statements, the audio information seems to be used by Google to send notifications to users’ phones.
Violation of California’s Privacy Act
Consequently, a class action complaint was filed against Google LLC and Alphabet Inc in California last Friday. The plaintiffs accuse Google of falsely claiming that devices do not record or process conversations unless a specific activation phrase is used. The complaint also states that Google falsely leads consumers to believe that their conversations or other audio are not recorded and/or sent to Google without their consent. Both these accusations, if true, are a violation of California’s Privacy Act (CCPA). Towards the end of July, Google Home users began reporting that Google was sending them alerts about information that could only have been obtained via their Google Home devices. One user on a Reddit forum, for example, wrote that they burned something in the kitchen, and subsequently received a notification from Google warning them that the smoke alarm had gone off. Another user on the same forum stated that their Google Home had picked up the sound of breaking glass and that they then received a notification of that event. Coincidentally, on 3 August, 2020, Google announced a strategic partnership with ADT, a leading US security and home automation provider. As part of this partnership, Google bought a 6.6% stake, worth $450 million, in the company. “Together with ADT, we look forward to delivering a smarter, more secure and helpful home”, Google states in their blog post.
