Meta's AI Glasses Have a Recording Problem. A Blinking Light Isn't the Whole Fix.

Meta's smart glasses have a small LED that's supposed to light up whenever the camera is recording, so people nearby know they might be on camera. Over the past few weeks, three separate stories have landed that, together, say a lot about the gap between "there's an indicator" and "people are actually protected."

The light was never hard to defeat

In August, CBS News reported on a woman who discovered a video of herself online that she never knew was filmed — taken by someone wearing Meta's glasses. It turned out some users had figured out how to block or physically tamper with the LED, then start recording, bypassing the one signal bystanders are supposed to be able to rely on.

Meta's fix, announced this week, changes the failure mode: previously, someone could cover the light before starting to record and the camera would work anyway. Now, covering the light after recording starts shuts the camera off. Meta says services that offered LED-tampering workarounds have since shut down as a result. That's a real improvement for the specific bypass that was being exploited — but it's a patch to a physical indicator, not a change to what happens to footage once someone has, in fact, consented to being on camera.

The lawsuit that just got bigger

On the same week as the LED fix, an amended complaint in an existing lawsuit against Meta was reported by Fortune, expanding the case to include bystanders who never wore or bought the glasses at all — people who were simply recorded by someone else's pair. The core argument is worth sitting with: even a bystander who notices the glasses and doesn't object to being filmed hasn't necessarily agreed to what happens next. If the footage gets processed by the glasses' AI features, the complaint alleges that imagery can be sent to Meta's servers and potentially used to help train its models — a use nobody standing in frame had any way to consent to, whether or not they noticed the light.

And the light itself may not be permanent

A separate report, citing leaked internal plans, described Meta testing a future "always aware" mode for next-generation glasses — background capture of audio and images throughout the day, so a wearer could later ask things like where they left their keys or what was discussed in an earlier conversation. Meta has publicly defended the recording light as an important safeguard even while this reporting suggests some internal exploration of a mode that wouldn't rely on active, per-clip recording the same way. Nothing here is confirmed as a shipped feature, but it's a reminder that the indicator light is a policy choice a company can revisit, not a physical constant of how the hardware works.

Why this is the harder version of a familiar problem

Most of what we write about here is a person deciding what happens to their own recording. This is the harder case: a camera that isn't yours, worn by someone else, pointed at you, where you have no settings screen at all — just a light you're trusting to be honest, on a device whose behavior can change with a company's next software update.

More from Offgrid Studio

If you're curious what a camera built the other way round looks like — one you set up on purpose, with footage that never leaves your own WiFi network — that's the idea behind OffgridCam, currently in TestFlight.