Meta's "Pervert Glasses" Scandal: 7 Million Sold, Privacy Debate Going Mainstream

Meta Ray-Ban Meta glasses are now a mainstream product

Seven million pairs in a single year is not a gadget niche. It is enough for the controversy to follow a product that is already on people's faces.

In 2025, Meta and EssilorLuxottica sold over 7 million AI glasses, after 2 million units so far had sold from launch through early 2025. That puts total sales at roughly 9 million units to date. With Meta and EssilorLuxottica also double production to meet demand, the category is clearly moving beyond early adopter curiosity.

That is why the scandal matters. When a product is this visible, privacy concerns stop being abstract and become a real brand and legal risk. Meta's glasses are estimated to account for more than 80% of all AI or smart glasses sales, so the controversy is attaching to category leadership rather than some obscure startup. Bulls will say that shows adoption is working. Bears will say the same scale makes any privacy misstep far more dangerous.

The privacy problem is the product design, not just bad users

The core issue is no longer whether these glasses can record. It is that the form factor makes discreet capture unusually easy. Meta says it is updating second-generation glasses so the camera shuts off if it detects the recording LED has been tampered with or destroyed; the camera already disables if the LED is covered. But that safeguard only helps if the warning is visible, credible, and hard to bypass.

Why the LED warning may not be enough

Meta's case is straightforward: a blinking LED is a visual warning, while a sound loud enough for nearby people to hear would not be practical in a glasses design. That keeps the product social and discreet in public.

The counterargument is that glasses do not need to be pointed like a phone. A smartphone usually has to be aimed directly at someone to record them; glasses can capture simply by looking at them. In that context, the visual cue matters more, not less. If bystanders struggle to notice a small temple LED in daylight, the warning can feel more symbolic than protective.

Human review adds a second privacy pressure point

Reports say Meta has been sending sensitive recordings to outsourced companies to review and annotate, while former Meta employees have described worker access to captured content in blunt terms such as "We See Everything".

That creates a second divide. Supporters can argue that third-party review is an operational necessity if the AI pipeline is being scaled toward around 20 million pairs a year by the end of 2026. Critics will say the moment sensitive media leaves the device and reaches outside processors, the privacy risk stops being only about one user's behavior and becomes a platform-design issue.

Key points to watch: - Whether tamper detection actually blocks covert use in real-world conditions. - Whether people can reliably see the LED warning, especially in daylight. - Whether Meta narrows data handling around annotation in a way that rebuilds trust.

The real bet is timing: scale fast before norms and rules harden

This is as much a timing battle as a privacy debate. Meta is pushing toward around 20 million pairs a year by the end of 2026, a scale path that could lower unit costs and strengthen market position before Apple, Google, or Samsung press harder. The scandal does not automatically end that thesis; it changes the window.

If adoption keeps outrunning restriction, Meta may be able to use scale and ecosystem depth to absorb the controversy. If culture or regulation pushes for a stronger recording warning or tighter data controls, the product becomes less discreet and the growth case gets harder to defend.