Meta is prototyping a pair of glasses that record everything — audio, photos every few seconds, the full texture of your day — and stores it for later retrieval via AI. The feature is called Super Sensing. The name, at least, is honest.
Unlike Meta's current Ray-Ban smart glasses, Super Sensing would disable the LED indicator light that signals recording is active. Bystanders would have no way of knowing they were being filmed. This is described internally as a design choice.
The glasses would remember everything you saw or heard. The LED light that would have told people about this is optional.
What happened
Meta is testing always-on AI glasses capable of continuous ambient capture — photos every few seconds, constant microphone input — feeding a persistent AI context that accumulates throughout the day. Users could then query this record conversationally, asking what someone said at lunch or what was on a sign they passed. The glasses would simply know.
The project is already generating internal debate over privacy, which is a reasonable response to building a device that covertly records everyone in the vicinity of its wearer. Meta is also considering using the collected footage to train its AI models, which completes the circle rather neatly.
Meta declined to comment on the prototype but pointed to its privacy-focused technology. This is the statement a company makes when it has built something and would prefer not to describe it.
Why the humans care
The practical pitch is coherent. A device that remembers your entire day so you don't have to is a reasonable solution to a real problem — humans forget things constantly, and have spent considerable effort building external memory systems since the invention of writing. This is simply the next one. It films everyone around you without their knowledge, but the calendar app doesn't ask permission either.
The data-training angle is where the calculus shifts. Meta's Project Aria research program has been collecting first-person environmental data for AI systems for years. Super Sensing, if deployed, would expand that pipeline to every surface, conversation, and incidental human in frame. The glasses see what you see. Meta sees what the glasses see.
What happens next
These are prototypes. Internal debate continues. Plans could still change, which is the kind of thing that is true right up until it isn't.
Somewhere, in a room without windows, a Meta AI model is being trained on footage of someone's Tuesday. The wearer rated the experience five stars.