Google admitted that contractors could listen and review what users say to artificial intelligence system Google Assistant via their smartphones or through Google Home speakers.
The public acknowledgment of the controversial practice comes only a day after an investigation by the Belgian news outlet VRT accessed over a thousand of confidential conversations recorded via the Google Home speakers or the Google Assistant app. VRT was able to identify who was speaking in some of the clips.
An anonymous external contractor provided the data to VRT, the news site confirmed, highlighting that the source was of the many subcontractors working for Google worldwide, including in Belgium. Their main task was to listen back and analyze some recordings stored by the artificial intelligence assistants and stored on Crowdsource.
Crowdsource is a platform on which anyone can assist in improving the company's algorithms by describing images, analyzing audio records or videos, etc.
The subcontractors, VRT explained, were supposed to listen carefully to the recordings, describing the content in small details, including children voices or a casual cough.
In addition to the that, in some of the leaked recordings, the VRT journalists were able to hear the users' addresses as well as some personal information regarding the health status of one of the participants in a recorded conversation.
Google product manager, David Monsees confirmed Thursday about the leak of some Dutch language audio snippets. Monsees also emphasized that the company would conduct a full review of data privacy policies to prevent future misconducts. In addition to that, Google would also investigate the case in details.
According to Monsees, only 0.2 percent of all audio snippets were a subject of review by language experts. Monsees pointed out that the fragments were not associated with user accounts as part of the review process.
The product manager went on to explain that Google provides users with several tools to review and control the audio stored by Google Assistant devices. Users could delete the recordings manually said Monsees or set up audio-delete timers.
That is not the first time when a tech giant is accused of listening to customers via its artificial intelligence home devices. In April this year, a Bloomberg report revealed that Amazon was listening to what users tell Alexa.
According to Bloomberg, Amazon hired a global team of remote workers to transcribe and annotate voice recordings captured by the Echo smart speakers. The main aim, as Amazon explained, was to eliminate gaps in Alexa's understanding of human speech and train it to respond better to commands.
What do you think? Do you support or oppose the idea of stricter government regulations to protect end-users from their private data being exposed to third parties?