Meta Workers Viewed Sensitive Ray-Ban Smart Glasses Footage, Report Says

Meta is facing renewed scrutiny over its approach to user privacy after a Swedish report found that employees of a Meta subcontractor viewed sensitive videos captured by Ray-Ban Meta smart glasses, including footage of people in private settings such as bathrooms.

The report, published in February through a collaboration of Swedish newspapers Svenska Dagbladet and Göteborgs-Posten, along with Kenyan freelance journalist Naipanoi Lepapa, is based on interviews with more than 30 current and former workers at Sama, a Kenya-headquartered data annotation company contracted by Meta. Some interviewees worked directly with video, image, and speech annotation for Meta's artificial intelligence (AI) systems. Others reported experience on unrelated projects, providing wider insight into annotation practices.

The workers described routinely viewing unfiltered footage recorded by users of Ray-Ban Meta smart glasses, raising critical issues about user consent and the adequacy of safeguards to prevent exposure of private or sensitive content. Data annotation often involves humans reviewing, labeling, and categorizing data—an essential process for training machine learning models that power facial recognition, object detection, and other AI features. The need for substantial human input in annotating and verifying this data underscores persistent privacy risks for consumers who use AI-powered devices.

Neither the journalists nor external observers were given access to the actual materials handled by Sama’s data annotators, nor to the areas where annotation took place. Despite this, several interviewees confirmed that annotation tasks occasionally included moments unintentionally captured in private circumstances, further highlighting potential flaws in data collection procedures.

Meta’s Ray-Ban smart glasses, developed with EssilorLuxottica, are marketed as hands-free, camera-equipped wearables powered by Meta AI. The glasses allow users to record and share first-person videos. They have drawn interest for their convenience, but also skepticism about their privacy implications—especially in Europe, where data protection is a priority and the General Data Protection Regulation (GDPR) enforces strict controls over personal data.

While Meta says it takes user privacy seriously and implements safeguards to protect sensitive information, this latest report intensifies calls from privacy advocates and regulators for stronger protections, more transparency, and oversight when AI systems rely on real-world footage. The situation also highlights the global supply chains behind AI products, wherein data handled by workers in countries like Kenya can have direct privacy implications for users worldwide.

The report did not provide evidence of direct GDPR violations, but it does shine a light on potential risk areas for AI model training and underscores the difficulty of ensuring responsible data handling across international vendors. As the use of wearable AI products expands, incidents like this may influence regulatory discussions and expectations around responsible AI development.

Source: arstechnica.com