Google must stop listening to voice recordings from people using its Assistant software in the European Union, German regulators said Thursday. The ban will last three months while Germany's data watchdog agency investigates privacy concerns.
The search giant's Assistant service, similar to Amazon's Alexa and Apple's Siri, lets people use their voice to turn on a light or get the morning's headlines through devices including a Google Home smart speaker, Android phone or iPhone. A small percentage of those voice recordings are heard by human workers at Google and third parties, with the goal of improving the software and better understanding people's queries, the company has said.
But those recordings could include sensitive and private information, not only from people using the device, but also from others in the same household, the Hamburg Commissioner for Data Protection and Freedom of Information said in a statement.
Last month, Google confirmed that third-party workers who analyze language data from the Assistant leaked private Dutch conversations. Belgian public broadcaster VRT NWS said more than 1,000 files had been leaked, including recordings from instances where users accidentally triggered Google's software.
"The use of language assistance systems must be done in a transparent way, so that an informed consent of the users is possible," Johannes Caspar, the Hamburg Commissioner for Data Protection and Freedom of Information, said in a statement. "In particular, this involves providing sufficient information and transparently informing those affected about the processing of voice commands, but also about the frequency and risks of mal-activation."
Google said it "paused" language reviews of the audio data after the Dutch audio leak. The company also said it's working with the German regulators.
"We are in touch with the Hamburg data protection authority and are assessing how we conduct audio reviews and help our users understand how data is used," a spokeswoman said in a statement. "These reviews help make voice recognition systems more inclusive of different accents and dialects across languages. We don't associate audio clips with user accounts during the review process, and only perform reviews for around 0.2% of all clips."
Other tech giants have been under fire for how they use their voice data. Amazon has said an "extremely small" number of Alexa recordings are annotated to help make its speech recognition systems better. Apple contractors reportedly hear private recordings too, including medical information, according to a report last month from The Guardian. Apple told The Guardian a "small portion" of the data is used to help improve Siri and dictation.
Originally published Aug. 1, 12:19 p.m. PT.
Update, 2:53 p.m.: Adds comment from Google.