Posted on Leave a comment

Meta smart glasses update tackles covert recording

meta smart glasses gen 2

Meta is rolling out a firmware update to its AI-powered smart glasses that shuts down the camera whenever the front-facing “capture” LED is covered or tampered with, closing a loophole that allowed users to film people without any visible warning. The move follows weeks of backlash and viral stories about nonconsensual recordings made with Meta’s glasses, which put the company’s latest wearable squarely in the crosshairs of regulators, privacy advocates, and anyone who doesn’t want to star in a stranger’s training data.

All of Meta’s current smart-glasses lines – including its second-generation Ray-Ban models and other partnerships – ship with a white capture LED that blinks once for a photo and continuously while recording video, and cannot be switched off by the wearer. Until now, Meta’s software was designed to block recording if the glasses detected that the LED was covered before a user hit the capture button, but crafty owners figured out they could start recording first and then slap tape or a sticker over the light to hide it. The new update goes further by disabling the camera if the system detects that the LED has been blocked, modified, or even physically destroyed mid-recording, turning a sneaky hack into a dead end for covert video.

Meta is also stepping up enforcement around LED tampering, removing ads and marketplace listings that offer modifications to disable or obscure the light and pursuing legal action against sellers who promise “stealth mode” recording. The company has framed the change as an industry-first safety feature, pointing out that traditional cameras and smartphones don’t forcibly shut down when someone covers their indicators. But while that sounds like a win for transparency, critics note that the update protects the integrity of the LED more than it protects bystanders, who still have no real way to refuse being filmed, demand deletion, or ensure their faces won’t be fed into AI systems.

Those concerns aren’t hypothetical. The UK Information Commissioner’s Office has already written to Meta over “concerning” reports about its AI smart glasses, explicitly raising questions about how recordings of unsuspecting people are handled and what safeguards exist for those who never agreed to be captured at all. Investigations and reporting have highlighted how footage from Meta’s glasses can end up in the hands of outsourced human reviewers, who have reportedly seen highly sensitive recordings – from financial information to intimate moments – while annotating clips for AI training. Meta has responded by stressing that media “stays on the user’s device” unless they choose to share it, but that doesn’t alleviate fears about how often users opt in to sharing and what happens once content enters the company’s broader AI ecosystem.

In parallel, news outlets have documented real-world cases where people discovered they were being filmed by Meta’s smart glasses without noticing the tiny LED, or only realized what happened after the fact. That feeds into a broader anxiety: these wearables look close enough to regular fashion eyewear that it can be hard to tell if you’re in front of a passive gadget or a sensor-laden camera rig quietly capturing your every move. For tech and AR enthusiasts, the ability to seamlessly record and query the world with AI is part of the appeal, but for everyone else, it starts to feel like a crowd-sourced surveillance network strapped to people’s faces.

From a geek-gadget perspective, Meta’s latest patch is a meaningful improvement: if you’re filming, the capture LED has to be visible and working, and crude hardware hacks to erase that signal now come with both technical roadblocks and potential policy consequences. Yet the core social problem remains unchanged – there is still no way for bystanders to truly opt out, and no simple mechanism for them to audit where their image goes once it’s recorded. As regulators circle and public pressure mounts, the LED update looks less like a final fix and more like an opening salvo in a bigger battle over what “always-on” AI hardware should be allowed to see, remember, and share about the people who never chose to wear it.

Our Sponsors

Geeks talk back