### Harvard Students Sound Alarm Over Privacy Risks of Meta Smart Glasses

Two Harvard students have demonstrated the potentially invasive capabilities of Meta smart glasses, revealing how they can be combined with face image search technology to uncover sensitive personal information from strangers in mere moments. AnhPhu Nguyen and Caine Ardayfio showed that by integrating Meta Ray Bans 2 with the PimEyes face search engine, one can identify a person’s name, address, and phone number simply by looking at them.

Nguyen and Ardayfio elaborated on their project in a Google document, explaining how this hybrid technology can quickly extract and compile an individual’s data using a large language model (LLM). This allows for automatic and comprehensive data gathering that would traditionally require much more effort and time.

Their inventive yet alarming creation, dubbed I-XRAY, leverages advances in LLMs to perform what was previously a tedious task of searching public databases. The students’ purpose was to draw attention to serious privacy issues in today’s rapidly evolving technological landscape.

The choice of Meta Ray Bans 2 was deliberate, as these glasses closely resemble standard eyewear, making them ideal for clandestine operations. During testing, unsuspecting individuals in public spaces were easily identified, highlighting the potential for misuse, such as tracking someone's home address. Some test subjects were told they were acquainted with the students, a ruse enabled by the detailed information obtained through the glasses.

Nguyen and Ardayfio stressed that the project is a cautionary tale, aiming to encourage people to opt out of invasive search engines like PimEyes. They made it clear they will not release the code for I-XRAY, but provided steps for others to protect their privacy from similar technologies. For instance, opting out of reverse face search engines like PimEyes can prevent identification attempts.

Despite raising awareness, real-world application showed mixed results. Not all test subjects were accurately identified, and some privacy risks were downplayed by companies like Meta and PimEyes. Nevertheless, the students’ work underscores the pressing need for public vigilance regarding personal data security.

With privacy regulations differing globally, such as the European Union’s stringent General Data Protection Regulation, U.S. citizens remain particularly vulnerable to the misuse of this kind of technology. Nguyen and Ardayfio hope that their project will promote greater awareness and proactive measures to safeguard personal information amidst the digital age’s rapid advancements.