A joint investigation by Svenska Dagbladet and Göteborgs-Posten went to Nairobi, Kenya and talked to the human data annotators – the real human workers who label and review footage from Meta Ray-Ban glasses to train the AI. What they found is… you guessed it, super bad! Update: March 4th, 12:36pm ET Meta emailed me back! See the comment/quote below & Update 2: March 4th, 4:02pm ET, Sama also emailed me! Added below.
Workers at Sama, one of Meta’s annotation subcontractors, describe reviewing video of people undressing, coming out of bathrooms naked, watching porn, having sex, and exposing bank card details. It’s all captured by someone wearing the glasses, often without the subject knowing they were on camera. One worker said: “We see everything — from living rooms to naked bodies.” Another noted there are no phones allowed in the office because the material is so sensitive it could trigger “enormous scandals” if anyone knew. Worth noting the company Sama, the Meta subcontractor is a Certified B Corp and part of the Clinton Global Initiative. How?
The anonymization that’s supposed to blur faces? Doesn’t work. “The algorithms sometimes miss. Especially in difficult lighting conditions, certain faces and bodies become visible,” a former Meta U.S. employee said.
Store staff selling the glasses in Sweden don’t know what gets shared. Multiple employees told reporters “everything stays locally in the app” but reporters ran packet analysis on the network traffic and found constant communication with Meta servers in Sweden and Denmark.
Meta’s response after two months of requests? A form letter pointing to their privacy policy. Sama didn’t respond at all, so I will try… I went to sama.com/contact-us and sent this article in asking for comment, I’ll @ them on the socials too. Update! March 4th, 12:36pm ET Meta emailed me back!
“Ray-Ban Meta glasses help you use AI, hands free, to answer questions about the world around you. Unless users choose to share media they’ve captured with Meta or others, that media stays on the user’s device. When people share content with Meta AI, we sometimes use contractors to review this data for the purpose of improving people’s experience, as many other companies do. We take steps to filter this data to protect people’s privacy and to help prevent identifying information from being reviewed.” – a Meta spokesperson
And Sama got back to me too, that is 2 for 2 –
“We do not comment on specific client relationships or projects, but Sama is a provider of data services for various tech companies to help them improve the accuracy and reliability of their AI models – including data annotation and labelling. Across this work, we maintain the highest standards of data security and privacy. Sama is compliant with international regulations including GDPR and CCPA, and we operate under rigorously audited policies and procedures designed to protect all customer information, including personally identifiable information (PII). This work is conducted in secure, access-controlled facilities. Personal devices are not permitted on production floors, and all team members undergo background checks and receive ongoing training in data protection, confidentiality, and responsible AI practices. Our teams receive living wages and full benefits, and have access to comprehensive wellness resources and on-site support.”– statement from Sama.
Data protection lawyer Kleanthi Sardeli at NOYB (who is actively reviewing the glasses) says the transparency problem is clear under GDPR. Kenya doesn’t have an EU adequacy decision for data protection, and an agreement isn’t expected anytime soon.
The full investigation is here… She Came Out of the Bathroom Naked, Employee Says — SvD/GP.
Here’s the thing, it’s not that I don’t like smart glasses… there is a lot of potential and innovation, just not from this guy. I think at this point, Facebook/Meta, and specifically Mark Zuckerberg, should not be involved in any way. Zuck has personally proven over and over again that his approach violates laws internationally, gets fined for it, and then does it again over and over again. What will it take? Why make your employees go through your messed-up approach, Mark? Sending it all off to Kenya just means you don’t even have the tech cooked enough. You are trying to Mechanical Turk your way onto very valuable real estate on people’s faces, and you’ve proven you cannot be trusted every single time. You’ve had multiple chances to get help, or get someone who will tell you straight when you are asking people to do things that are illegal. Stop it, dude.
On the privacy page for the glasses it says, in bold “Designed for privacy, controlled by you” Not true. And “You’re in control of your data and content.” Not true. Expect another lawsuit, another settlement that is a penny for Meta.
Every year we see Mark talking to U.S. politicians about something he did, and the result is nothing. Maybe a fine. Well… how about instead of saying AI needs a six-month pause, we say “Mark can’t work on recording glasses for six months”… something to send a signal that he keeps lying and causing other people a lot of problems.
Here is Vanity Fair pointing this out with a quote from Mark, from “The Facebook Effect” by David Kirkpatrick….
“I have over 4,000 e-mails, pictures, addresses, SNS,” Zuckerberg bragged to a friend. “People just submitted it. I don’t know why. They ‘trust me.’ Dumb fucks.”