The Meta Ray-Ban smart glasses are currently among the most advanced smart glasses money can buy in the market. With a second-generation model already on sale, Meta claims to have sold over 7 million units in 2025 alone. The glasses integrate AI capabilities for real-time environmental analysis and recording, featuring perks like voice-activated AI assistance, live translation, and hands-free capture. However, a recent investigation has uncovered a disturbing behind-the-scenes reality – offshore contractors annotating footage for AI training are routinely exposed to highly sensitive and intimate user videos, including private moments users likely never intended to share.
According to reports from Swedish outlets Svenska Dagbladet and Göteborgs-Posten, data labelers working for Meta’s partner, Sama, in Nairobi, Kenya, described reviewing footage that captures users in bathrooms, undressing, engaging in sexual activity, filming bank cards or credit details, watching pornography, or recording explicit scenes.
A Meta spokesperson, however, has come up with a response to the matter, citing the AI Terms of Service and Privacy Policy, which allows the company to use multimedia content for AI training, once they give consent. The user consent has to be given in order to use the Meta AI services, which many do without reading.
One annotator recounted a specific clip where a man set the glasses on a bedside table and left the room, only for his wife to enter and change clothes. The footage was captured without apparent awareness from either party.
The contractors expressed significant discomfort, noting they feel compelled to annotate such material or risk losing their jobs. Meta’s instructions reportedly prohibit questioning the content, leaving workers to process deeply personal glimpses into strangers’ lives.
One employee remarked, “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. You are not supposed to question it. If you start asking questions, you are gone.” Another added, “I don’t think they know, because if they knew they wouldn’t be recording.”
“You think that if they knew about the extent of the data collection, no one would dare to use the glasses,” another annotator told the newspapers.
How Meta Ray-Ban smart glasses use your data for AI training
The Meta Ray-Ban glasses feature built-in cameras and microphones that record first-person perspectives when users activate AI features or recording modes. To improve the AI’s understanding of visual contexts, raw footage is sent to servers and then outsourced to human annotators in low-cost regions like Kenya, Colombia, and India, which is a common practice across the AI industry for training models on real-world data.
Meta’s AI terms of service state that it allows the company to review user interactions, including media and conversations, manually or automatically, and warn users against sharing sensitive information. However, once users opt into AI functionalities, it requires agreeing to data sharing, after which users effectively lose control over how footage is used or by whom it is viewed.
What does Meta say
Meta has not yet provided a detailed substantive reply to the investigation. However, a company spokesperson stated, “When live AI is being used, we process that media according to the Meta AI Terms of Service and Privacy Policy.”
The revelations from the report bring out concerns related to user data privacy. Since the device features a camera and mic, both of which can be hidden easily by malicious actors, concerns of stealth recording leading to the harassment of an individual emerge. The case also brings up the exploitative nature of offshore data labeling jobs, where workers in developing countries have to handle traumatic or intimate content similar to social media moderation horrors.