Apple Siri, Google Assistant pause audio “grading” amidst privacy concerns

JC Torres - Aug 2, 2019, 7:04 am CDT
2
Apple Siri, Google Assistant pause audio “grading” amidst privacy concerns

It may have been a running joke that Google and other companies are hearing your embarrassing conversations or expressions said in the presence of smart speakers and smart assistants but little did we know that reality was far stranger than fiction. It turned out, strangers are actually privy to some, but not all, of those audio clips for the usual sake of improving the service. Governments and regulators, however, are putting that process under a microscope, causing Apple and Google to halt such systems in Europe while they try to evaluate privacy implications and, if possible, work around the legal requirements.

Apple formally calls it a “grading” process but it’s a common system that Google and Amazon share. All three, unsurprisingly, have been named in different exposes regarding how that system can be used and abused. In a nutshell, these Big Tech companies send a supposedly anonymized and random snippet of audio to human auditors, often contractors, to evaluate and tag if the smart assistant interpreted the audio correctly.

The reality is that those audio clips could actually contain information that might be embarrassing at least or identifiable at worst. Considering how these assistants sometimes get triggered accidentally, itself a flaw of the software, there is an off chance something unintended gets recorded and sent to those strangers without the user’s knowledge, much less consent.

Responding to the reports, both Google and Apple have reportedly put their respective programs on hold while they review things and probably hope that the scrutiny dies down. CNBC reports that Google has scheduled a three-month hiatus in Europe while Apple is halting it worldwide.

It’s not a permanent end to the problem, of course. It’s unlikely that Big Tech will stop employing this method as human ears are sometimes still needed to verify AI interpretations. Apple, for its part, will make the program voluntary and will ask users first rather than assuming they already agreed to the process under the more general terms of service. Amazon, the other Big Tech named in the reports, has not yet made a similar move.


Must Read Bits & Bytes