Walking around Meta Connect 2026, everyone's sporting smart glasses in all sorts of shapes, colours and sizes. It's a marked difference here, a tech bubble where 'pervert glasses' are not a concern.
Meta has ditched the camera on its newest smart glasses, opting for audio-only Ray-Bans. Privacy controls to address the 'perv glasses' backlash are noticeably missing. Outside Connect, the public backlash against wearable surveillance tech is gaining steam, so it's incredibly curious timing that this year, the company is launching Ray-Bans with no camera for recording others.
Audio-only glasses have been years in the making, according to Alex Himel, Meta's vice president of wearables. 'Three years ago, it wasn’t clear if AI glasses were going to be a thing. Back then, every other company was backing out of the space,' Himel says. 'Then two years ago, we saw that audio was the top use case, the biggest potential for AI, and that it should work well in a pair of audio glasses.'
The audio-only models start at 43 grams – heavier than normal wire-frame glasses, but noticeably lighter than the roughly 50g Ray-Ban Meta Gen 2 and the 69g Meta Ray-Ban Display. The arms are thinner. They come in two styles: the Mad Men-esque Clubmaster and the more Wayfarer-reminiscent Burbank. The hinges extend for wider faces; the temple arms and nose pads are adjustable. There are 23 colour and lens combinations, and they can support the same wider prescription range as the Ray-Ban Meta Optics lineup. Even more so than the standard Meta Ray-Bans, these look and wear like normal glasses. And with a £349 starting price, they're cheaper too.
Free of a power-hungry camera, the battery lasts an estimated 12 hours on a single charge. The most technical things you can do with them are interact with Meta AI, take phone calls, and listen to podcasts or music. And that's it. 'They don’t record,' Himel says. 'The only thing they do, which is the same thing as an Alexa device in your house or some other devices, they’re just listening for a specific wake word. They’re not sending anything… If they hear [the wake word], then they start listening to what you’re saying. When you’re done talking, then it’s processing, and that’s visible to the AI. It gets processed, and that’s kind of it.'
