In a new report, Bloomberg claims Facebook recruited human contractors to transcribe users’ voice chats on the Messenger app to check the interpretation accuracy of the company’s artificial intelligence.
Facebook then confirmed to the press that it had ceased the exercise. The company emphasized it only reviewed audio samples from Messenger users who agreed to participate in an opt-in program launched in 2015.
However, Bloomberg says that Facebook’s current data-use policy failed to clearly inform its users that third parties might listen to their audio files.
Meanwhile, the social media giant also refrained from disclosing the true nature of those audio samples supplied to human transcribers, causing concerns of unethical practices among those employees.
TaskUS, one of Facebook’s contractors responsible for carrying out user conversion reviews, moderations, and ads screenings, doesn’t allow its employees to mention their clients’ real names. Instead, Bloomberg says those workers refer to them by the code name “Prism.”
That said, Facebook was also not the only tech company that faced public backlashes for audio collection and the potential invasion of personal privacy. MobileSyrup reported back in late July that Apple contractors routinely listened to sensitive user information transmitted through Siri and similar stories have come out regarding Microsoft’s Cortana, Amazon’s Alexa and Google Assistant.
However, unless tech companies become more transparent about their data collection practices, abandoning human-involved review programs may mean nothing more than a token gesture.