Meta's Ray‑Ban cameras catch workers watching bathroom footage, staff say

Meta’s Ray‑Ban “AI Glasses” have become the focus of a privacy‑related class‑action claim after a Swedish‑Kenyan investigative report alleged that subcontractor staff were regularly viewing footage that captured intimate moments inside users’ homes. The report, a joint effort by Svenska Dagbladet, Göteborgs‑Posten and Kenyan freelance journalist Naipanoi Lepapa, is based on interviews with more than 30 employees of Sama, the Kenya‑based firm that provides data‑annotation services for Meta’s wearable devices. According to the translated interviews, several annotators said they had been asked to review video clips showing people having sex or emerging from the bathroom naked, with one employee recounting, “I saw a video where a man puts the glasses on the bedside table and leaves the room. Shortly afterwards, his wife comes in and changes her clothes.” (Ars Technica)

Sama workers described the material as “privacy‑sensitive data that is fed straight into the tech giant’s systems,” a flow that made many uncomfortable. The journalists noted that they were not granted access to the actual annotation workstations or the raw data, but the consistency of the testimonies suggests a systematic practice rather than isolated incidents. Former U.S. Meta employees, also interviewed for the piece, corroborated the existence of live‑annotation pipelines for multiple Meta projects, indicating that the glasses are part of a broader data‑collection ecosystem that extends beyond the wearable itself. (Ars Technica)

Meta’s response, delivered to the BBC, confirmed that user‑generated content from the glasses is sometimes shared with contractors “to improve people’s experience, as many other companies do.” The company emphasized that the data is “first filtered to protect people’s privacy,” citing automatic face‑blurring as an example of its safeguards. Meta’s privacy policy for the wearables states that photos and videos are uploaded to the cloud only when users enable “cloud processing” or interact with the Meta AI service, and that users can disable this feature in the device settings. The policy also notes that livestream audio, video, and text transcripts are processed by “machine learning and trained reviewers” and may be shared with third‑party vendors. (Ars Technica)

Legal analysts have not yet weighed in on the viability of the proposed class action, but the allegations raise questions about the adequacy of Meta’s de‑identification methods. While face‑blurring can obscure visual identifiers, it does not eliminate contextual cues—such as the layout of a bathroom or the presence of a partner—that can still reveal a user’s identity. Moreover, the report suggests that annotators were expected to continue their work despite recognizing the invasive nature of the content, a practice that could run afoul of labor‑rights standards in addition to privacy law. The combination of alleged “concealing the facts” by Meta and the discomfort expressed by Sama staff could amplify regulatory scrutiny, especially in jurisdictions that have recently tightened rules on biometric and location‑based data. (Ars Technica)

The episode arrives at a moment when Meta is pushing the Ray‑Ban glasses as a flagship product in the emerging wearables market, a narrative echoed in more optimistic coverage from The Verge and TechCrunch. Those outlets highlight the glasses’ sleek design and integration with Meta’s AI services, but they have not addressed the privacy controversy. As the class‑action suit gains momentum, the company may be forced to reassess its data‑handling pipeline, potentially revising the consent mechanisms that currently allow cloud processing by default. Until then, users of the AI glasses face a stark choice: continue leveraging the device’s capabilities while trusting Meta’s internal filters, or opt out of cloud‑based features altogether to avoid becoming part of a data stream that, according to multiple employee accounts, sometimes includes footage from the most private corners of their lives. (Ars Technica)