Students Used Meta’s Smart Glasses to Dox People Live

Two Harvard students have developed a demonstration showcasing the capability of smart glasses to utilize facial recognition technology to instantly access personal information such as identities, phone numbers, and addresses. This demonstration, which leverages current, widely available technology like Ray-Ban Meta smart glasses and public databases, has raised significant concerns regarding privacy.

AnhPhu Nguyen, one of the creators, released a video demonstrating the technology in action, which was subsequently highlighted by 404 Media. Named I-XRAY, the technology operates by using the Meta smart glasses’ live streaming capability to Instagram. This video stream is monitored by a computer program that utilizes artificial intelligence to identify faces. The identified facial images are then cross-referenced with public databases to retrieve names, addresses, phone numbers, and even relatives. This information is then relayed back through a phone application.

In the demonstration, Nguyen and the co-developer, Caine Ardayfio, are shown using the glasses to identify several classmates, along with their addresses and relatives, in real-time. The video also depicts the students engaging with strangers on public transit, using information obtained via the tech to converse as though they were acquaintances.

Facial recognition technology has been known to deliver highly accurate results for years, and the I-XRAY tool primarily integrates a variety of existing technologies. Part of its functionality is based on PimEyes, a facial recognition search engine noted for its precision and accessibility, as reported by The New York Times in 2022. Notably, concern over this technology has been amplifying, especially after Clearview AI’s application in aiding law enforcement. Nguyen and Ardayfio’s innovation stands out for pairing this technology with a consumer gadget that is discreet and widely accessible.

Nguyen and Ardayfio clarified in a document explaining the project that their objective is not to misuse the technology and they do not intend to release it. Instead, their goal is to highlight that these capabilities are feasible with existing technology. They emphasize that I-XRAY’s uniqueness lies in its use of large language models (LLMs) to automatically associate names and photos from extensive data sources.

Privacy has long been a contentious issue with smart glasses. Google Glass’s initial failure was partly due to public backlash against being recorded without consent. In contrast, the past decade has seen an increased acceptance of being filmed, driven by smartphones, vloggers, and TikTok. However, contemporary smart glasses, like the Ray-Ban Meta, are less conspicuous than Google Glass was, which raises additional privacy concerns.

The Ray-Ban Meta glasses used in this demonstration resemble typical Ray-Bans, making them less noticeable as recording devices. While this design is crucial for smart glasses adoption, it complicates the detection of active cameras. Although these glasses are equipped with a privacy light indicating video recording, it often goes unnoticed in bright daylight and crowded spaces.

Meta advises users via its privacy policy for the Ray-Bans to respect others and clearly indicate when recording, livestreaming, or taking photos. Nonetheless, adherence to this wearable etiquette, as Meta suggests, is not guaranteed by all users. Meta was contacted for further comment by The Verge but did not immediately respond.

This scenario highlights the potential misuse of smart glasses, though there are measures individuals can take to protect themselves. Nguyen and Ardayfio provide a list of reverse face search and people search databases that offer opt-out options in their document. However, it is nearly impossible to completely erase one’s online presence; the best one can do is to make their information less accessible.

Source link