Meta has started disabling the camera function of its smart glasses
Meta has started disabling the camera function on thousands of smart glasses after detecting that users had manipulated, damaged or deliberately disabled the recording LED.
The LED is a central part of the glasses’ privacy design. It is intended to alert people nearby when a photo or video is being taken. If manipulation is detected, the camera remains unavailable until the LED works again. Where the damage is irreversible, the loss of the camera function is permanent.
Meta estimates that fewer than 0.1% of its smart glasses have been manipulated in this way. With millions of devices sold, that still amounts to several thousand affected glasses. The company has also removed advertisements for services offering to disable the LED and threatened legal action against the providers.
From a privacy perspective, this is significant.
Smart glasses make recording technology far less visible. A smartphone being held up usually gives people some indication that they may be filmed. A camera built into an ordinary-looking pair of glasses is much harder to recognise. That makes the recording LED a practical transparency mechanism. Once it is covered or disabled, people nearby may have no realistic opportunity to understand that their image or behaviour is being recorded.
Meta is now also closing a further loophole. The camera should stop recording if the LED is covered during an ongoing recording, rather than checking its visibility only when the recording begins.
Technical safeguards of this kind are welcome. Their limitations are equally clear. A functioning LED only helps if people notice it and understand what it means. Privacy by design therefore needs more than a small light. It requires safeguards that anticipate deliberate misuse, clear information for affected individuals and effective action when recordings are shared unlawfully.
Meta has also stated that facial recognition is currently not something it is working on for the glasses. Given how unobtrusive the devices already are, adding real-time identification would raise the privacy risks considerably.