Meta's AI Glasses Reportedly Send Sensitive Footage to Human Reviewers in Kenya
An investigation by two Swedish newspapers reveals that contractors in Nairobi review footage captured by Meta's smart glasses — including content from bathrooms and intimate moments.
Summary
Meta's AI-powered smart glasses may be sending sensitive footage to human reviewers in Nairobi, Kenya, according to an investigation by Swedish outlets *Svenska Dagbladet* and *Göteborgs-Posten*, published last week. According to the report, Nairobi-based contractors have seen videos captured with the glasses that include "bathroom visits, sex and other intimate moments." One worker quoted by *Svenska Dagbladet* states: "We see everything — from living rooms to naked bodies."
The reviewers in question are AI annotators — workers who label images, text, or audio to help AI systems make sense of the data they are training on. Meta, in response to *The Verge*, confirms the practice exists but frames it as standard industry procedure.
In practice
Meta spokesperson Tracy Clayton told *The Verge* that media captured by the smart glasses "stays on the user's device" unless users choose to share it with other people or with Meta. The company adds that it "sometimes uses contractors to review this data for the purpose of improving people's experience, as many other companies do," and that it "takes steps to filter this data to protect people's privacy and to help prevent identifying information from being reviewed."
However, workers in Kenya partially contradict this account: a former Meta employee reportedly told *Svenska Dagbladet* that faces in annotation data are blurred automatically, but Nairobi workers say this "does not always work as intended" and some faces remain visible. Another person cited in the investigation reports that users' bank cards are sometimes visible in the footage being reviewed.
Beyond images, the Kenyan reviewers also work with transcriptions, verifying that the glasses' AI assistant provides correct answers to questions users ask aloud.
Meta's Ray-Ban and Oakley camera-equipped smart glasses have grown significantly in popularity: according to data cited in the article, EssilorLuxottica — Meta's partner in developing the glasses — sold over 7 million units in 2025, more than tripling combined sales from 2023 and 2024. Last year, Meta updated its privacy policy to keep Meta AI with camera use enabled by default "unless you turn off 'Hey Meta'," and also stopped allowing wearers to opt out of storing their voice recordings in the cloud.
Who benefits / who loses
The investigation has prompted at least one proposed class action lawsuit in the United States, accusing Meta of violating false advertising and privacy laws. The complaint specifically cites Meta's claim that its glasses were designed to protect privacy, arguing the company "hid the alarming reality: that use of the AI features results in a stranger halfway around the world watching the most private moments of a person's life."
The UK's Information Commissioner's Office has questioned Meta about the claims in *Svenska Dagbladet*'s reporting. Privacy advocates have also raised concerns about Meta's alleged plans to integrate facial recognition into its smart glasses, with the Electronic Privacy Information Center describing the prospect as a "grave risk to privacy, safety, and civil liberties."
From a broader perspective, the situation is compounded by the fact that people filmed are not necessarily the glasses' wearers — they may be third parties captured involuntarily, with no knowledge or consent that their images could be reviewed by human contractors.
Why it matters
- The practice of human review of footage captured by wearable devices raises privacy questions that go beyond typical privacy policies: users are rarely aware that images from their daily lives may be seen by third-party contractors hired to train AI systems.
- Reported failures in automatic face-blurring and the visibility of bank cards suggest that technical protection mechanisms are not robust enough for the volume and sensitivity of content generated by always-on camera glasses.
- With more than 7 million units sold in 2025 alone, the scale of the product makes the implications of this practice significantly broader than they would be for a niche device.
- The investigation arrives as Meta is expanding the glasses' capabilities — including potential facial recognition — which intensifies the regulatory and judicial scrutiny already underway.