Amazon team members speaking under the conditions of anonymity have now come forward to reveal that Alexa Data Services employees listening in on Alexa interactions are granted access to customer location data, Bloomberg reports. Specifically, the recordings that the employees listen to are associated with longitude and latitude data, corresponding to the user's location when the interaction occurred.
Teams that have access to that are spread across India, Costa Rica, the U.S., and several other countries and it isn't immediately clear whether any have actually used that data. Amazon has responded by stating that access is severely limited and controlled. Its policies are designed to be strict and prevent employees from misusing the information as well as to ensure that audits occur regularly to check for misuse.
The anonymous employees agree with Amazon’s statements but claim that most of the work to make personal user data such as phone numbers, information from their contacts, location tracking based on their smartphone, home or work addresses, and more were only recently restricted. For some workers in the Alexa Data Services team, restrictions weren't in place until after news broke that employees had, in fact, been listening.
The news is in complete opposition to claims that the company made after it was determined that human workers were being given access to Alexa recordings. The move was allowed by a clause tucked deep within the terms and conditions for the service and associated devices. The team, comprised of AI engineers and associated experts, was said to be auditing the interactions in order to make improvements and creating new ones based on what consumers are asking for.
Users need to explicitly opt out of the program to avoid the possibility of having their recordings or recording transcripts viewed by employees.
The discovery raised concerns about how well a user's personal information and privacy could be maintained if workers were able to just listen in whenever they interacted with the technology. Amazon had said at the time that not only are recordings only taken when the user starts an interaction but that that the information was effectively anonymized.
Namely, it said that employees who audit recorded interactions with Echo or third-party Alexa-enabled devices had no access to identifying information about the customers. The statement also ignored the faults in Alexa's hot word detection and glitches that have previously led to undesired listening or responses from the AI.
The latest reports seem to indicate that isn't entirely true since the information that is being provided is more than enough to remove any sense of privacy via a simple Google Maps search.
This could have a widespread impact
There is legislation already under consideration to begin shifting the rules away from allowing the underlying activity here that allows companies to take user data wholesale. That's geared toward reducing the collection of data specifically from teenagers and younger users as well as addressing and keeping an eye on collection policies via audits.
The bipartisan effort may ultimately be undermined by the increasing number of reports and growing concern about data ownership and privacy. While Amazon's current practices appear to be relatively limited in scope, AI is continuously being incorporated into more IoT technology and with a growing level of functionality.
Amazon's iteration of the technology represents a top competitor in that space from a company that many consumers have come to trust. As such, news of the company's direct involvement in shady practices and apparent dishonesty about its own practices will undoubtedly impact the entire IoT and AI industry. That could lead to diminishing marketability and slowing growth in what might otherwise be one of the most promising technologies to emerge in recent times.