UK regulator probes Meta over intimate smart glasses videos

Walking around your house with smart glasses feels futuristic until you realize someone half a world away might be watching the replay.

A joint investigation by Swedish papers Svenska Dagbladet and Göteborgs-Posten just exposed a messy reality behind the Ray-Ban Meta glasses: outsourced contractors in Kenya have been reviewing deeply personal videos captured by wearers, including footage of people using the bathroom, watching porn, and having sex. "We see everything, from living rooms to naked bodies," one worker told reporters.

Inside the annotation pipeline

The review work was handled by Sama, an outsourcing firm in Nairobi hired to label training data and check whether Meta AI answered user queries correctly.

Meta claims that recordings stay on the device unless people deliberately share them with Meta AI to ask questions, and that contractors only see media to improve the experience. Meta also insists that privacy filters blur out faces before humans lay eyes on anything. But the contractors told a very different story, saying the automated face-blurring routinely failed.

In one instance, a worker described how a man left his glasses sitting on a bedroom dresser while recording, inadvertently capturing his wife undressing. Sama defended its operations, saying production floors are strictly monitored, personal phones are banned, and workers get background checks. But high-security cubicles don't change the fact that real humans were watching intimate home videos.

Regulators circle as a class action lands

The fallout has been immediate. The UK Information Commissioner's Office (ICO) announced it is writing to Meta to demand explanations on how the company complies with data protection laws, stating that smart devices must give users clear control and total transparency.

Across the Atlantic, Meta is already facing a fresh class-action lawsuit. Filed by Clarkson Law Firm on behalf of two consumers in California and New Jersey, the complaint accuses Meta and eyewear partner Luxottica of deceptive marketing.

The suit points out that Meta plastered ads with slogans like "designed for privacy" and "built for your privacy," while burying the reality of human review inside sprawling terms of service. With over seven million pairs of Meta smart glasses sold in 2025 alone, the lawyers argue millions of buyers had their sensitive footage funneled into data pipelines with no clear way to opt out.

Why it matters

AI hardware has a dirty secret, and Meta just dragged it into the daylight: frontier multimodality still runs on cheap human labor.

If you own a pair of smart glasses, this is your wake-up call. Many buyers genuinely believed their queries were processed solely by machine learning models on a server. Instead, asking Meta AI "what am I looking at?" effectively gave an overseas contractor permission to review your camera feed. Right now, taking these devices into private spaces like bedrooms or bathrooms simply isn't worth the privacy risk.

For the broader tech industry, this controversy threatens the entire category of ambient wearables. We think Meta's core mistake was trying to have it both ways: marketing the device as an effortless consumer fashion accessory while relying on an aggressive, enterprise-style data collection pipeline behind the scenes.

Expect regulators in Europe and the US to crack down hard on vague terms of service disclosures. If hardware makers want consumers to adopt always-on cameras and microphones, they will likely be forced to offer strict on-device processing or unmistakable, explicit toggles before any human touches the footage.