In February, Meta contractors in Kenya told Swedish newspapers Svenska Dagbladet and Göteborgs-Posten that the company required them to review disturbing and sensitive footage captured by its Ray-Ban AI glasses.
Some reported seeing wearers naked or using the toilet. Another saw a man’s wife undressing in their bedroom, after he left the glasses on a table, the joint investigation found. Other footage they reviewed included entire “sex scenes.”
“You understand that it is someone’s private life you are looking at, but at the same time you are just expected to carry out the work,” one employee told the Swedish newspapers. “You are not supposed to question it. If you start asking questions, you are gone.”
It seems their hunch was correct.
Two months after the worker’s allegations were published in the newspapers, Meta responded in highly questionable fashion. It terminated its entire contract with the Kenyan company, Sama, the BBC reports — a decision that a Kenyan worker’s organization alleges was in retaliation to the workers speaking out.
Meta didn’t address that allegation, but told the BBC that it “decided to end our work with Sama because they don’t meet our standards,” while stressing that it took the worker’s claims “seriously.”
“Photos and videos are private to users,” a spokesperson said. “Humans review AI content to improve product performance, for which we get clear user consent.”
For its part, Sama defended its workers.
“Sama has consistently met the operational, security and quality standards required across our client engagements, including with Meta,” it said in a statement. “At no point were we notified of any failure to meet those standards, and we stand firmly behind the quality and integrity of our work.”
The allegations shine a light on the dark underbelly of the AI industry, and tech more broadly: how much of it is driven by underpaid workers overseas, and how much data it reviews behind the scenes. At Sama, the workers were performing data annotation, a process that involves manually labelling images, videos, and other content so an AI model knows what it’s looking at during training. For the Ray-Ban glasses, this is supposed to help their built-in AI function more seamlessly.
They’ll also add to the perception that Meta’s AI wearables are “pervert glasses,” allowing users to discreetly record people without their consent or knowledge. While the Ray-Bans have a light to indicate when they’re recording, these can reportedly be disabled, and many have discovered tricks to cover them. Sama workers reported that some of the users appeared unaware that their glasses were recording.
“People can record themselves in the wrong way and not even know what they are recording,” one told the Swedish newspapers.
The bombshell claims have put Meta under the microscope. The UK’s Information Commissioners Office contacted Meta about the “concerning” reports. And Kenya’s Office of the Data Protection Commissioner announced it would conduct an investigation into the Meta glasses’ potential privacy violations.
Naftali Wambalo of the Africa Tech Workers Movement said he has spoken with workers involved in the Meta glasses, alleging that Meta terminated its contract with Meta because it didn’t want workers speaking out.
“What I think are the standards they are talking about here are standards of secrecy,” Wambalo told the BBC.
More on AI: Meta Installing Software on Employee Computers to Track Everything They Do, Feed the Data to AI