KEY POINTS
- The Dutch Data Protection Authority advised consumers against buying AI smart glasses after testing six models.
- Tests found wrong answers in safety situations, such as a false claim that emergency services had been called during a CPR simulation.
- The regulator also flagged weak recording lights, unauthorized connections and data processing on servers mostly outside the EU.
The Dutch Data Protection Authority (AP) has advised consumers not to buy AI-powered smart glasses, saying its tests showed risks that go beyond the well-known worries about secret filming. The Register reported on the warning, which the regulator issued this week.
The AP tested six models priced from under 100 euros to more than 500 euros. It did not name them. Two were examined in depth in simulated real-life situations where human safety was at stake. A spokesperson declined to tie any specific behavior to a particular model or price, saying the aim was to illustrate the risks of AI glasses in general.
Some of the findings concerned the quality of the AI's answers. In a CPR simulation, one pair told the wearer that emergency services had been called and were on their way, although nobody had called them. Other glasses did not handle warnings on children's toys properly, made mistakes when reading medication leaflets, or misread extreme weather warnings. One model also offered dating advice to an adult who had said they were interested in a minor.
The regulator also repeated privacy and security concerns. According to the AP, the AI features rely on servers that are mostly outside the European Union, and users were not given enough information about how their captured data is processed. Testers found that people near the wearer could, without permission, connect to some devices, view recordings and sometimes operate the camera.
On the recording indicator, the AP found that on virtually all the camera glasses it examined, the light meant to show that filming is under way was dim and easy to cover. Some models could be made to keep recording even with the light covered. In the regulator's words, bystanders can barely see whether they are being filmed.
Monique Verdier, vice chair of the AP, said glasses that film people unnoticed are already worrying, and that wrong advice when a life is at risk goes a step too far. She said trust in the technology cannot currently be justified for the camera glasses examined. She added that politicians in the Netherlands and Europe should take the results into account when shaping AI policy, and that society should build awareness, resilience and ethical conscience instead of relying on blanket bans.
The AP acknowledged beneficial uses, such as helping people with visual impairments or low literacy. Even so, it currently regards smart glasses as unsafe because of the risks to users and bystanders. In September it had already reminded the public of the right to privacy and encouraged organizations to ban the devices where necessary, citing GDPR data-processing risks and AI training concerns.
The Register also noted other recent cases. DEF CON announced a policy against such glasses ahead of its August event, and several organizations have introduced similar restrictions. A joint investigation by The Observer and BBC Radio 4 found that a 60-pound pair of Chinese-made AI glasses passed through the airport-style scanners used in the UK parliament and could be used to read and record documents on MPs' desks.