Reference no: EM132867608
In April 2019, news site Bloomberg broke the story that Amazon, Apple and Google all "employ staff who listen to voice recordings from their smart speakers and voice assistant apps." The fact that the machines have the capability to listen shouldn't come as a shock, since they wouldn't be of much use if they didn't. "Hey Google!" wouldn't do much without the ability to get a response. What seems to get overlooked is the assumption that the recording ends after the question or task has been completed.
The terms and conditions for all three companies make it clear that recordings are used to improve the accuracy of their speech recognition programs. What isn't specified (though this will likely change after this revelation) is that recordings are sampled by human beings, not computer software. Voice clips are transcribed and annotated only with an account number and the serial number of the device used with no further personal details.
Google has been here before, facing a backlash when it admitted that Gmail messages were scanned for keywords to show you relevant advertisements on your Gmail home page. Subsequent clarification that the scanning was done by artificial intelligence (AI) rather than real people helped to reassure users, but the assumption by Google corporate that users wouldn't have an issue with the idea of their e-mails being scanned revealed a comfort level with technological capability that far exceeded awareness of user preferences.
Amazon responded to the growing media coverage of the story with the following statement:
"We have strict technical and operational safeguards, and have a zero-tolerance policy for the abuse of our system. Employees do not have direct access to information that can identify the person or account as part of this workflow."
Reading more like a pre-emptive legal assertion than a genuine response to customer concerns, Amazon's statement carefully avoided any reference to specific details in the Bloomberg story such as the allegation that voice clip reviewers frequently shared amusing clips with one another in an internal chat room, or that one clip was so distressing that the reviewer suspected a sexual assault may have taken place.
Hearing or Listening?
The technology of these digital personal assistants continues to develop at a rapid pace. The smart speakers in particular are becoming increasingly interconnected with the technology in your home such as appliances, lighting and air conditioning systems, and domestic security systems. They are technically always "hearing" for the "wake word" needed to activate them. According to brand, "Alexa!," "OK Google," or "Hey Siri!," will activate the devices and they will begin listening to your voice commands.
Security researchers have offered reassurances that it would be relatively easily to identify if entire conversations were being recorded and sent back for analysis (as opposed to short random clips) and that, so far, no evidence has been found of that happening. However, when you compare Amazon's statement to the allegations made in the Bloomberg story, it's easy to see why skepticism remains high.
Technology journalists such as Leonid Bershidsky, who wrote an opinion piece to accompany the Bloomberg story, argue that these revelations should not come as a surprise
Why did the Bloomberg piece gain so much attention?