The social networking site explained that it used the records of only those users who chose the option to decrypt audio files in the messenger, and thus checked how well the neural network works. But as experts point out, the problem is that Facebook doesn't mention audio recordings in the data-use policy and users didn't provide their consent to the company to transcribe their communications and share them with third-party firms.
“This is not a correct behaviour, the overall procedural lack of transparency for the end-users. Regarding the purpose of the transcription announced by Facebook, I believe it could be realistic”, Pierluigi Paganini, CTO at Cybaze SPA, explained.
As far as why Facebook decided to do this without user permission, the tech expert believes it may have to do with the fact that the Silicon Valley giant was secretly testing new features and attempted to keep its technology secret.
“This kind of technology could be also abused for surveillance purposes, [its] likely Facebook avoided revealing it to avoid being in a new media storm”, he noted. He added that besides supposedly testing neural networks, like the company claimed, the purpose could be to train AI “to transcribe every conversation and then perform any kind of analysis on the overall conversations recorded over the time”.
Fong Choong Fook, the CEO of Malaysia-based LGMS, a cyber-security firm, echoed these views:
“It is plausible that end user audio messages were transcribed to train Artificial Intelligence systems. Inevitably, when training AI systems, large among of data input is typically required”.
Fook believes that Facebook might have 'sanitised' the user data first prior to feeding it into AI training systems, explaining that this process is aimed at anonymising the names and any potential private information in the data, but it's not clear how much data remains.
“Unfortunately, to whatever extent Facebook sanitization works, it remains something unknown to its users and authorities. This could be the very factor that may lead to potentially wild speculations, given the tainted history of how Facebook has handled personal user data in recent years”, the cyber security firm CEO said.
Back in March 2018, Facebook let the personal data of around 87 million of its users be harvested by a data analysis firm - Cambridge Analytica - without their permission through a special app. The information was allegedly used to help target political advertising. In July this year the US Federal Trade Commission issued Facebook a $5 billion penalty over the data breach scandal.
But Munzir Ahmad, a New Delhi tech journalist and Founder of Sky Televentures, has said that Facebook is not the only big tech firm that plays “dirty tricks” on its users to make money. However, when the companies get caught, they tend to come up with “lame excuses” as to why they did it:
“Not only Facebook but the mighty Apple, who always criticised Facebook for data breach, was also listening to user’s conversation via Siri. If these companies are so true, then why did they stop listening to voice recordings after expose? If they are true and their purpose is genuine, then they should not have stopped it. I mean, if you are not thief then there’s no reason to be scared. Right? Look at the Apple, Facebook, Google and Amazon, they all seemed scared when this particular news came. This proves they are not doing these things for users benefits at all”, the journalist stressed, also calling these activities illegal.
Views and opinions expressed in this article are those of the speakers and do not necessarily reflect those of Sputnik.
The views and opinions expressed in the article do not necessarily reflect those of Sputnik.