Two Harvard students have created a chilling demonstration of how smart glasses equipped with facial recognition technology can instantly reveal people’s identities, phone numbers, and addresses. The demo, which uses current, widely available tech like the Ray-Ban Meta smart glasses and public databases, showcases just how accessible this kind of doxxing can be.

AnhPhu Nguyen, one of the students behind the project, shared a video of the technology in action that was later highlighted by 404 Media. Named I-XRAY, the tech utilizes Meta’s smart glasses to livestream video to Instagram, where a computer program tracks the stream and employs AI to recognize faces. These images are then cross-referenced with public databases to extract names, addresses, phone numbers, and even information about relatives, which is then displayed in a phone app.

In the video, Nguyen and his partner, Caine Ardayfio, demonstrate how the glasses can identify several classmates, their addresses, and their relatives’ names in real time. Even more unsettling, the students are shown interacting with strangers on public transit, pretending to know them based on information gathered through the tech.

Facial recognition technology has long been disturbingly precise, and I-XRAY essentially combines existing tech in a new way. Part of its functionality relies on PimEyes, a facial recognition tool described by The New York Times as an “alarmingly accurate” face search engine. The controversy around this technology has grown, especially since Clearview AI began using facial recognition to aid law enforcement. What’s unique about I-XRAY is its integration with a consumer gadget that is both discreet and accessible.

“The purpose of building this tool is not for misuse, and we are not releasing it,” Nguyen and Ardayfio clarify in a document outlining the project. Instead, they aim to highlight that this isn’t a distant, dystopian possibility but rather something achievable now with existing technology. They emphasize that I-XRAY is novel in how it leverages large language models (LLMs) to automatically draw connections between names and images from extensive data sources.

Privacy concerns have always surrounded smart glasses. Google Glass, for instance, faced backlash for enabling people to record others without their consent in public spaces. However, in the years since, society has become more accustomed to being filmed due to smartphones, vlogging, and platforms like TikTok. What’s unnerving about modern smart glasses is their inconspicuous design.

The Ray-Ban Meta glasses used in the demo resemble ordinary Ray-Bans, making it difficult for people to tell when they might be recorded. Although these glasses have a privacy light that turns on when recording, it is easy to miss in bright outdoor settings or crowded areas.

Meta advises users to respect people’s privacy and clearly signal when capturing video, livestreaming, or taking photos. Yet, individuals may not adhere to these guidelines, regardless of Meta’s recommendations. When contacted for comment, Meta did not provide an immediate response.

This demonstration serves as a stark reminder of how easily smart glasses can be misused, but Nguyen and Ardayfio suggest some steps to protect personal information. They provide a list of reverse face search engines and people search databases that allow opting out. Nonetheless, it’s important to recognize that completely removing your digital footprint is nearly impossible; you can only make your data less accessible.