Meta secretly integrates facial recognition into smart Ray-Ban glasses
Meta secretly integrated facial recognition code into the Ray-Ban glasses app. The hidden system creates biometric profiles of people in plain sight. Data protection experts warn of significant surveillance risks.
Secret return of facial recognition
In recent months, Meta has quietly installed code for a facial recognition feature on millions of smartphones. The software components are hidden in the Meta-AI app, which is required as a companion for the group’s Ray-Ban smart glasses. The company is apparently preparing a technology that will identify people in the field of vision of the glasses in real time. There is currently no official announcement for end customers. The function, known internally as NameTag, uses three local AI models that work directly on the end device. The first model recognizes faces in the environment, the second precisely cuts them out of the image material. A third model finally converts the images into biometric feature profiles, so-called faceprints. The system then compares the data with a database on the smartphone. Carriers should be notified immediately of any matches.
Code analysis reveals new details
Like US Magazine Wired reported in an analysis of the source code, the core components of the system have been integrated into the app since January 2026. Independent security experts confirmed that the function is almost ready for use. Recent updates internally renamed the tool “Connections”. The user interface already asks users to remember people they have met. According to the investigation, the app initially saves unknown faces in a separate folder that is marked as pending. It remains unclear where the comparison database for recognizing faces in everyday life will ultimately come from. Such a system could be helpful in everyday life for people with visual impairments. Data protection advocates, on the other hand, warn against a gradual normalization of mass surveillance. Real-time detection in public spaces poses significant risks of abuse, for example through targeted stalking. The technology deeply intrudes on the privacy of passers-by without consent.
Meta denies concrete plans
The development is surprising because Meta actually discontinued its facial recognition system on Facebook in 2021 after massive criticism. The software previously automatically assigned users names on uploaded photos. At that time, the company deleted over a billion faceprints that had already been created and promised more protection for privacy. Previous legal disputes over unlawful data storage cost the company a lot of money. The violations resulted in massive fines, including a $650 million settlement in the US state of Illinois and another $1.4 billion settlement in Texas. A spokesman for Meta explained that the technology is currently only being researched and that they are not building a central facial database. So far, no function has been activated for end customers. However, the secret integration into the software shows that the project is already very advanced in the background.
