Meta's Always-On Smart Glasses Raise Serious Privacy Questions

Meta is reportedly developing a new generation of smart glasses that would record video continuously, rather than requiring the wearer to tap a button or speak a command first. According to reporting from The Verge, the project is internally referred to as involving 'super sensing' capabilities, with an AI layer processing the footage in real time to give the wearer contextual information about their surroundings. If that sounds familiar, it is because Meta's Ray-Ban smart glasses already caused a stir when Harvard students demonstrated last year that you could use the glasses' camera combined with facial recognition to identify strangers on the street within seconds. Now Meta appears to be leaning into exactly that direction, rather than away from it.

The distinction between 'press to record' and 'always recording' is not a small one. It is arguably the single most important design decision a wearable camera company can make, and it sits at the heart of a genuine ethical debate that the tech industry has been avoiding for years. When Google Glass launched in 2013, the backlash against 'Glassholes' was swift and social. People did not want to be filmed by strangers without knowing it. Nothing fundamental about human psychology has changed since then, even if our tolerance for surveillance has quietly crept upward.

What 'Always On' Actually Means

There are two plausible architectures for an always-recording device. The first is full continuous capture, where footage is stored locally or streamed to the cloud, similar to a dashcam. The second is a rolling buffer model, where the glasses constantly record but only retain the last 30 to 60 seconds unless the user actively saves a clip. Apple reportedly uses something similar with its rumoured camera-equipped Apple Watch concepts. The rolling buffer approach is arguably less invasive because nothing is permanently saved without intention, but it still means the camera is active in every conversation, in every bathroom, at every medical appointment.

Meta has not confirmed the project or its technical specifications. The Verge's sources describe it as still in development, with no launch timeline attached. But given that Meta has already sold millions of Ray-Ban smart glasses and has been aggressively expanding its wearables line, the jump to always-on recording is clearly something the company is seriously exploring rather than idly sketching.

From a purely technical standpoint, always-on recording creates enormous challenges around battery life, storage, and on-device processing. Running a camera continuously is power-hungry. Compressing and storing or transmitting that footage in real time requires serious compute. The fact that Meta is reportedly pursuing this suggests the company believes it has solutions to those problems, possibly through dedicated AI chips in the frame itself.

The Consent Problem Nobody Has Solved

The bigger issue is consent, and it is one the entire industry has failed to adequately address. When you wear always-recording glasses into a coffee shop, you are capturing the barista, the person at the next table, the child in the pushchair by the window. None of them agreed to be filmed. None of them will know they were. In the European Union, GDPR creates real legal exposure here, and several member states have already flagged smart glasses as a potential compliance nightmare. In the United States, the legal landscape is patchwork at best, with some states requiring all-party consent for recording and others not.

Meta's answer to this concern with the Ray-Bans was a small LED indicator light that glows when recording. In practice, almost nobody notices it. An always-on version would presumably keep that light on constantly, at which point it stops functioning as a meaningful signal and becomes background noise.

There is a version of this technology that could be genuinely useful. Hands-free navigation, real-time translation of signs and menus, assistance for people with visual impairments or memory conditions. Those are real applications with real benefits. But delivering them does not require recording every human being in the wearer's field of vision all day long. The question worth asking is whether Meta is building this because the use cases demand always-on recording, or because always-on recording is what makes the data valuable to Meta's advertising and AI training pipelines. The answer to that question should shape how regulators respond before these glasses reach shop shelves.