The Dutch privacy watchdog has advised people not to buy smart glasses after tests found some devices gave dangerously wrong information, including false claims that an ambulance had been called.
The Autoriteit Persoonsgegevens (AP) tested six pairs of glasses sold in the Netherlands. It said the devices misread medicine instructions, mishandled severe weather warnings and, in one test, gave dating advice to an adult who indicated a romantic interest in a minor. The AP did not identify the brands tested.
In a simulated emergency, one pair repeatedly claimed an ambulance was on its way after a request for CPR advice, although emergency services had not been contacted. The watchdog stopped short of calling for the glasses to be banned from sale.
The AP said the tested glasses cost from under €100 (£84) to over €500 and came from different continents. It also raised concerns that recording indicator lights could be obscured, that some devices had security flaws allowing nearby people to access recordings or operate cameras remotely, and that data was often exported outside the EU.
The investigation focused primarily on algorithms and artificial intelligence, rather than data protection. The AP said smart glasses can be valuable to people who need help understanding their surroundings, but argued that reliable AI is especially important for those users.
Privacy concerns around smart glasses have also included covert filming. Meta, a leading seller, recently released a camera-free version; it is unknown whether Meta glasses were included in the AP’s tests. The watchdog told the BBC it would not name the products.
The BBC said it had approached Meta, TCL and HTC for comment. Meta has previously highlighted how its glasses can help people with visual impairments by using AI to describe the wearer’s surroundings or read text.
In February, reports said outsourced workers were able to view sensitive content recorded on Meta smart glasses. Meta said workers might sometimes review video and images to help improve its AI, and that it took data protection seriously and would filter content to blur faces.
