Meta released a mandatory software update this week for its Ray-Ban AI glasses. If the small white light on the front of the glasses, the one that blinks during recording, is physically removed or destroyed, the camera shuts off entirely. The company called it an industry first, and technically, it is.
The update was necessary because removing that light had turned into a small business. Paid services to strip the LED from the glasses were being advertised openly, including on Meta's own marketplace. In October 2025, the University of San Francisco warned students after a man was caught filming women with modified glasses and posting the footage. A woman in London was filmed while shopping and later pressured to pay money to have the footage taken down. A BBC investigation found dozens of influencers across TikTok and Instagram doing exactly the same thing. One woman's clip hit 1.3 million views and included her phone number.
So the fix addresses a real problem. The issue is scope. Meta shipped 7.4 million units in 2025, a 281% increase from the year before, and holds roughly 84% of the global smart glasses market. The tamper-detection update only applies to second-generation devices going forward. Every first-generation pair already sold, millions of them, still records normally if the light is covered with something as simple as a sticker. Those stickers are available on Amazon India right now, one set marketed specifically for discreet recording in business meetings.
There is also the question of what happens when the camera is working exactly as designed. Most bystanders either miss the blinking light or assume it is a Bluetooth indicator. A person sitting across from someone wearing these glasses has no practical way to refuse being recorded, no way to know whether footage has been uploaded, and no way to ask for it to be deleted. Researchers at the Cognitive Privacy Project reported last month that most people have no idea they are being filmed even when the light is functioning perfectly.
When users do share footage with Meta AI, that footage is sometimes reviewed by human contractors. A Swedish newspaper investigation in early 2026 found workers at a Nairobi subcontractor viewing intimate footage from inside users' homes: people undressing, visiting bathrooms, having sex. One documented case involved a woman who walked in front of glasses left on a bedside table. That footage was sent overseas for labeling without her knowledge. A class action lawsuit followed.
On the same day Meta announced the privacy fix, it also announced that its AI can now use anyone's public Instagram photos to create AI-generated images, unless you manually opt out. The pattern is consistent: a safety measure on one side, a data expansion on the other.
The bigger issue sits just below the surface. In June 2026, researchers at the Electronic Frontier Foundation found dormant facial recognition code already deployed to more than 50 million Ray-Ban glasses through a routine app update. The feature, internally called "Name Tag," would scan faces in real time and match them against a database. It is not live for users yet. The code is already on the device. An internal Meta document from 2025, reported by the New York Times, described plans to launch the feature during a period when civil society groups would have their attention elsewhere. A coalition of 75 organizations including the ACLU has since written to Meta calling the feature a line that should not be crossed.
For anyone who runs a business with employees, customers, or clients who interact in person, this matters in a practical way. Camera glasses that look identical to prescription eyewear are already in millions of pockets. The rules governing what someone can record in your offices, your shop, or your meeting rooms were written before any of this existed. Most legal frameworks have no clear answer for what a bystander can do. Checking your own country's workplace recording laws is not a bad use of an hour.