MENLO PARK — For two years, Meta has been quietly building a surveillance controversy around a pair of sunglasses. In June, it shipped a facial recognition feature in its smart glasses app without telling users. Before that, a pair of Harvard students showed the world how easily those same glasses could identify strangers on the street. Meta apologized, delayed, patched, and pressed on. Now it wants consumers to strap a more powerful version onto their faces.
That pitch begins Wednesday, when Mark Zuckerberg takes the stage at 4 p.m. Pacific for Meta Connect 2026, a keynote at the company’s Menlo Park campus centered almost entirely on what Meta’s wearables team has been building since the Ray-Ban Meta became an unexpected bestseller. New models are coming. More capable sensors are coming. And the company most associated with “move fast and break things” is attempting both with the most intimate computing device ever to find mainstream adoption.
The stakes are higher than they look. Wearables are now Meta’s actual growth bet, not a side project, and Connect has quietly transformed from a VR showcase into a hardware event for glasses. But the question that nobody inside Menlo Park seems eager to answer in public is whether consumers who absorbed the privacy panic have shorter or longer memories than the product cycle. Meta’s own documentation of what its glasses can do to a stranger has made that question harder to sidestep than it was eighteen months ago.
Wednesday’s lineup includes a new generation of Ray-Ban Meta, the camera-embedded sunglasses that have already sold in the millions. Developers familiar with the keynote expect at least one prescription-compatible model, closing a gap that has kept the glasses out of daily rotation for roughly half the population. A second model, codenamed Phoenix internally, is rumored to carry a price between $1,000 and $2,000, targeting buyers who want something closer to a wearable computer than a fashion item. Meta has confirmed neither pricing nor specifications ahead of the event.
The more consequential reveal, if it materializes, is what Meta has been calling its “super sensing” tier. Codenames Aperol and Bellini have circulated among hardware developers for months. These models are described internally as glasses that can process ambient audio and video continuously, using AI to generate real-time context about the user’s environment without them ever touching a phone. In theory: you walk into a meeting and the glasses already know who is in the room. In practice: everyone nearby becomes subject to passive data collection they never consented to. The privacy implications are not subtle. They have, however, generated considerable developer interest.
Meta’s facial recognition feature for smart glasses, added to its app without flagging it as a new capability, drew criticism precisely because it made theoretical privacy concerns operational. The announcement that more powerful, always-on sensing is coming next does not land in isolation. It arrives on top of that documented history, and for many potential buyers it will require a more convincing explanation than the company has so far offered.
The camera-free “Luna” model, also expected Wednesday, reads almost as a strategic concession to that criticism. No camera, no passive video, just bone conduction audio and a small heads-up display. It is the product Meta can demonstrate without triggering a new round of regulatory inquiries. Engadget’s preview of the event suggests this model targets buyers who want the wearable form factor without the privacy exposure, which raises the question of whether that segment is large enough to justify its own product line.
Meta is also expected to announce expanded availability for its glasses in international markets currently blocked by pending regulatory approvals, including several in Asia and the Middle East where consumer demand has been suppressed rather than absent.
For developers, Wednesday doubles as a negotiation. The September 24 sessions will detail what the glasses SDK permits third-party applications to do with sensor data — the dividing line between accessory and platform. A permissive data policy with a rich SDK is the most commercially valuable outcome Meta could announce. It is also the scenario that would immediately accelerate scrutiny from regulators who have been watching this category with considerable attention to its surveillance implications.
Samsung showed its own AI-powered glasses at Unpacked earlier this summer, integrating Google’s Gemini directly with Snapdragon AR1 in a form factor targeting the same prosumer buyers Meta is chasing with Phoenix. The race to define what AI glasses actually do as a product category is still genuinely open, and no single company has yet built a device that works well enough, looks normal enough, and costs little enough to push the category into genuine mass adoption.
Meta holds advantages no competitor can quickly replicate: the Luxottica partnership, the Ray-Ban brand, and an installed base already running into the tens of millions. What it no longer holds is the benefit of the doubt from a broad consumer market. Wednesday’s keynote streams live beginning 4 p.m. Pacific at Meta’s Connect 2026 page, and it will be watched by people genuinely curious about the hardware, by regulators who want to understand what the sensors can collect, and by researchers who will have the new specifications analyzed before the first unit ships.
Zuckerberg has always been good at selling futures. At Connect 2026, he needs to sell the same future twice: once as a product, and once as something safe enough to wear.
