Meta can watch you through AI Smart Glasses

smart glasses spy on you

The advertisement is slick and global: Swedish ice hockey legend Peter Forsberg tries on a sleek pair of black glasses. “Who is Sweden’s greatest hockey player of all time?” he asks the glasses, which are not just any eyewear — they are Meta’s new AI-powered “Ray-Ban Glasses,” marketed as an all-in-one assistant for work, travel, photography, and real-time language translation. Meta, the company behind Facebook, Instagram, and WhatsApp, promises control over privacy. But the reality behind the product is far more complex — and troubling.

At the top of a hotel in Nairobi, under a grey, oppressive sky, a worker sits nervously. If his employer discovers he’s speaking to journalists, he could lose everything. He is one of thousands of “data annotators” in Kenya, subcontracted by Sama to train Meta’s AI systems. Their job? To watch, label, and analyze endless streams of video captured by Meta’s smart glasses — including footage from inside people’s homes, bedrooms, and even bathrooms.

“Sometimes you see someone going to the toilet, or getting undressed,” the worker says. “I don’t think they know. If they did, they wouldn’t be recording.”

The Human Engine of AI

In September 2025, Mark Zuckerberg unveiled the glasses at Meta’s headquarters in Menlo Park, California. On stage, he showcased their capabilities: live translation, facial recognition, and more. But behind the polished presentation lies a global supply chain of manual labor. In Nairobi, workers draw boxes around objects, label scenes, and transcribe conversations — all to make the AI “smarter.”

“Machine learning” is often portrayed as a purely technological process, but in reality, it is built on human hands — and often, in low-income countries.

Privacy at Risk

The annotators describe videos of intimate moments, people undressing, and even explicit content. “We see everything — from living rooms to naked bodies,” says one worker. “Meta has that type of content in its databases.”

They also report seeing sensitive financial information, like bank cards, and even people watching pornography while wearing the glasses. “If these clips were leaked, it would be an enormous scandal,” says another worker.

Meta’s privacy policy states that data may be reviewed by humans, but users are not explicitly told that their most private moments could be viewed by workers thousands of miles away.

What Do Users Really Know?

In Sweden, retailers reassure customers that they are in control of their data. “Nothing is shared with Meta,” one salesperson claims. But technical tests by journalists show otherwise: the glasses require an internet connection to function, and data is routinely sent to Meta’s servers in Sweden and Denmark.

“Once the material is fed into the models, the user in practice loses control over how it is used,” says Kleanthi Sardeli, a data protection lawyer with NOYB, a privacy advocacy group.

What Do Users Really Know?

In Sweden, retailers reassure customers that they are in control of their data. “Nothing is shared with Meta,” one salesperson claims. But technical tests by journalists show otherwise: the glasses require an internet connection to function, and data is routinely sent to Meta’s servers in Sweden and Denmark.

“Once the material is fed into the models, the user in practice loses control over how it is used,” says Kleanthi Sardeli, a data protection lawyer with NOYB, a privacy advocacy group.

The Human Behind the Algorithm

Back in Nairobi, the data annotators are bound by strict confidentiality agreements. They are not supposed to question what they see. “When you watch these videos, it feels like you’re looking straight into someone’s private life,” says one worker. “But it’s a job. You’re expected to do it.”

One annotator sums up the paradox: “If people knew the extent of the data collection, no one would dare to use these glasses.”

Meta’s Response

After repeated requests, Meta’s spokesperson sent a brief statement: “When live AI is being used, we process that media according to the Meta AI Terms of Service and Privacy Policy.” The company did not address specific questions about private content, subcontractor oversight, or how users are informed.

Sama, the subcontractor, did not respond to requests for comment.

Bottomline: Never trust Meta with anything.

Source: SvD

Leave a Reply

Your email address will not be published. Required fields are marked *