Amazon stores recordings of Alexa interactions and turns them over to internal staff and outside contractors for review
Bloomberg reporters learned that -- despite public pronouncements to the contrary -- Amazon has an "annotation team" of thousands of people all over the world, charged with reviewing recordings made by Alexa devices in the field, with both staffers and contractors listening to conversations that Alexa owners have had with and near their devices.
The annotation team's goal is to improve Alexa voice recognition and command parsing.
The team has sometimes encountered evidence of criminal conduct, such as a recording of a sexual assault, but were advised by their superiors at Amazon not to take action.
Amazon defends the practice by saying that only a small proportion of the material gathered by Alexa devices is reviewed by the team, and says that "strict technical and operational safeguards" and "a zero tolerance policy for the abuse of our system" mean that customers have no need to worry about privacy violations. The recordings the team reviews are tagged with the customer's first name, account number and the serial number of their device.
Recordings are made any time an Alexa device detects an activation word, like "Alexa" or "Echo" or "computer." Sometimes recordings are made inadvertently when these words are used in other contexts, or are misinterpreted by the devices.
Apple and Google also have human teams that review audio from devices, but with different privacy safeguards.
The team comprises a mix of contractors and full-time Amazon employees who work in outposts from Boston to Costa Rica, India and Romania, according to the people, who signed nondisclosure agreements barring them from speaking publicly about the program. They work nine hours a day, with each reviewer parsing as many as 1,000 audio clips per shift, according to two workers based at Amazon's Bucharest office, which takes up the top three floors of the Globalworth building in the Romanian capital's up-and-coming Pipera district. The modern facility stands out amid the crumbling infrastructure and bears no exterior sign advertising Amazon's presence.
The work is mostly mundane. One worker in Boston said he mined accumulated voice data for specific utterances such as "Taylor Swift" and annotated them to indicate the searcher meant the musical artist. Occasionally the listeners pick up things Echo owners likely would rather stay private: a woman singing badly off key in the shower, say, or a child screaming for help. The teams use internal chat rooms to share files when they need help parsing a muddled word-or come across an amusing recording.
Amazon Workers Are Listening to What You Tell Alexa [Matt Day, Giles Turner, and Natalia Drozdiak/Bloomberg]