Is Facebook always listening to me?
Company admits its contractors transcribed Messenger app voice recordings, but insists it has now stopped the practice
Facebook has admitted that it paid contractors to listen to voice recordings of users’ private conversations.
The social media giant confirmed that it had been “transcribing users’ audio” after company insiders told Bloomberg that people were hired to analyse conversations on the firm’s Messenger service.
The aim was to check the accuracy of speech-to-text artificial intelligence (AI) systems, the news site says.
Subscribe to The Week
Escape your echo chamber. Get the facts behind the news, plus analysis from multiple perspectives.
Sign up for The Week's Free Newsletters
From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.
From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.
Facebook told Bloomberg that it has now stopped transcribing user recordings after rival tech firms, including Apple, Google and Amazon, came under fire for a similar practice.
“Much like Apple and Google, we paused human review of audio more than a week ago,” a Facebook spokesperson said.
Although “human oversight” of automated systems is commonplace in the tech industry, The Guardian says that examining user conversations from chat apps and voice-activated speakers poses “a significantly higher level of concern over privacy”.
How - and why - Facebook records your voice
As reported by The Daily Telegraph, audio recordings were analysed courtesy of the “voice chat transcription option” in Facebook’s Messenger app. The option lets users record a brief audio clip, which is then sent to a recipient and transcribed into text through AI.
Facebook’s privacy policy doesn’t explicitly say that these recordings would be evaluated by a human moderator, but it does state that “content, communications and other information you provide” may be shared with “vendors and service providers who support our business”, the paper adds.
Facebook claims that it used human moderators to help improve the accuracy of its products, which include the AI system that transcribes audio messages into text on the Messenger app, the Financial Times notes.
The firm also insists that the identity of users was anonymised before their voice recordings were passed on to the contractors, the FT says.��
What did the contractors hear?
Sources familiar with the practice told Bloomberg that the contractors were not told how the audio files were obtained or where they were recorded. The moderators were simply asked to listen to the recordings and transcribe them.
The company insiders say the moderators could listen to “Facebook users’ conversations”, some of which contained “vulgar content”, the news site reports. The contractors, however, were unaware as to why Facebook needed the recordings transcribed.
Contractors working for other firms appear to have been exposed to more extreme content.
In an interview with Vice, one moderator, working for the Microsoft-owned video calling service Skype, said that they could hear conversations that could “clearly be described as phone sex”, as well as “people entering full addresses” or asking the Cortana AI system to “provide search returns on pornography queries.
“While I don’t know exactly what one could do with this information, it seems odd to me that it isn’t being handled in a more controlled environment,” the contractor said.
So is Facebook always listening to me?
Judging by comments made to Bloomberg, it would appear that Facebook does not listen to every user at all times - especially now the firm states that it no longer examines voice recordings.
But Jake Moore, a cybersecurity specialist at ESET, told Forbes that Facebook “already admits to scanning Messenger content such as text, images and links for many reasons so transcribing audio or video content is really just an add on to what they are already doing.
“Micro targeted advertising is their multimillion-dollar business so naturally there may be data within this audio which will add to people’s profile,” he said. “However, where this becomes a privacy matter or even grey area is, people wrongly tend to assume that this content is private from all eyes and ears.”
Create an account with the same email registered to your subscription to unlock access.
Sign up for Today's Best Articles in your inbox
A free daily email with the biggest news stories of the day – and the best features from TheWeek.com
-
Duchess of Gloucester: the hard-working royal you've never heard of
Under The Radar Outer royal 'never expected' to do duties but has stepped up to the plate
By Chas Newkey-Burden, The Week UK Published
-
Are 'judge shopping' rules a blow to Republicans?
Today's Big Question How the abortion pill case got to the Supreme Court
By Joel Mathis, The Week US Published
-
Climate change is driving Indian women to choose sterilization
under the radar Faced with losing their jobs, they are making a life-altering decision
By Theara Coleman, The Week US Published
-
How social media is limiting political content
The Explainer Critics say Meta's 'extraordinary move' to have less politics in users' feeds could be 'actively muzzling civic action'
By Chas Newkey-Burden, The Week UK Published
-
Cyberflashing, fake news and the new crimes in the Online Safety Act
The Explainer UK's first conviction demonstrates scope of controversial law that critics describe as a threat to privacy and free speech
By Harriet Marsden, The Week UK Last updated
-
TikTok predicts 'creative bravery' trends for 2024
talking point Will it be a banner year for the platform, or will it isolate its biggest audience?
By Theara Coleman, The Week US Published
-
Elon Musk's 'frivolous' but precedent-setting free speech fight with Media Matters
Talking Point The lawsuit is just the latest in Musk's ongoing tension with social media watchdogs
By Theara Coleman, The Week US Published
-
Twitter's year of Elon Musk: what happens next?
Why Everyone's Talking About 'Your platform is dying', says one commentator, but new CEO is aiming for profitability next year
By Chas Newkey-Burden, The Week UK Published
-
How AI might influence democracy in 2024
The Explainer Threat from bots and deepfakes stalks key elections around the world next year
By Harriet Marsden, The Week UK Last updated
-
Pros and cons of social media content moderation
Pros and Cons Where do you draw the line between online safety and freedom of speech?
By Theara Coleman Published
-
Turns out Facebook isn't as polarizing as previously thought
Talking Point New studies show that, contrary to prior belief, the algorithm has little effect on driving polarization
By Theara Coleman Published