Is users' privacy safe and duly protected when engaging with AI bots like Google Assistant and Amazon's Alexa? Apparently not as the recent voice data leak in Europe seemed to indicate the possibility of compromise due to human intervention and unintended recordings.
It turned out snippets of voice interactions by humans with Google Assistant have been accidentally recorded, stored and transcribed, and these processes were characterized by Google as "necessary to creating products like the Google Assistant."
In the same process, human workers or language experts employed by the tech giant were given access to users' private information. Unknowingly, users' sensitive data - names, addresses, and other personal details - were exposed to strangers, The Verge reported.
The same thing had transpired for Alexa users, whose unauthorized audio clips were also worked by human transcribers for the purpose of improving the service.
Per the same report, the cases brought to light were limited to Dutch and Flemish speakers who had regular interactions with the Google Assistant. Google had acknowledged this, but the company clarified only a small fraction of the recordings, estimated at 0.2 percent, have been processed for transcription by humans.
However, in the report on the matter aired in Belgium recently, one of the contractors had admitted that his transcription works led him to picture of real happenings on the human voices he listened to. The contractor also shared he detected physical violence in one of the recordings.
His testimony somehow indicated that the practices by Google and Amazon were inadvertently intruding into the private lives of users.
Privacy advocates have expressed alarm over the incident, saying it was wrong on the part of the tech companies that they failed to inform users that voice recordings, even if accidental, have been taking place. The same advocates fear that without prior knowledge, users are prone to reveal information to Google and Amazon that otherwise, they want to remain private.
In a statement published in a blog post, Google has admitted that the language reviewers in question have "violated our data security policies by leaking confidential Dutch audio data."
It should be noted that the breach happened as the company insisted that "a wide range of safeguards to protect user privacy throughout the entire review process," was in place.
Nonetheless, Google said that it is addressing the issue.
"Our Security and Privacy Response teams have been activated on this issue, are investigating and we will take action. We are conducting a full review of our safeguards in this space to prevent misconduct like this from happening again," David Monsees, Google Search product manager, wrote in the blog.