A video buried inside the macOS Tahoe 26.7 release candidate has given us our first real look at how Apple's long-rumored camera-equipped AirPods actually work. The clip, discovered by MacRumors contributor Aaron Perris and shared on X, shows a user wearing AirPods while holding up a book. The voiceover explains the premise: "With Visual Intelligence, your world becomes savable. See something you like? Just ask me to save it for later."

It is, apparently, exactly what it sounds like. The cameras embedded in the AirPods feed visual data to Siri, which then interprets what the user is looking at and logs it. The product carries the internal codename B790 and has been tracked by Bloomberg's Mark Gurman for months. References in macOS 26.7 include setup screens for Visual Intelligence on AirPods, plus an alert system that warns users if their hair is obstructing the lens.

A Different Kind of Sight

Here is where it gets interesting. The placement of cameras on the sides of a person's head, rather than facing forward from the chest or face, gives these AirPods a perspective more like that of a bird than a human. Most birds have eyes positioned laterally, providing what researchers call monocular vision. This setup allows some species to see nearly 340 degrees around them, watching for predators while simultaneously foraging. The tradeoff is diminished depth perception compared to forward-facing binocular eyes.

Advertisement

AirPods cameras will operate under similar constraints. Each earbud sees a slice of the world from a different angle. Neither lens faces directly forward in the user's line of sight. The cameras will almost certainly be low-resolution infrared sensors, not the high-megapixel modules you find on smartphones. According to analyst Ming-Chi Kuo, they may be similar to what Apple already uses for Face ID. This is intentional. Apple is building a tool for situational AI awareness, not a conventional imaging device.

Why Dormant Mode Makes Sense

The leaked video provides a subtle clue about power management. The demonstration appears to show the user issuing a voice command before the system responds. This suggests the cameras operate in a dormant state, only activating on-demand. There is good reason for this. Imaging hardware consumes significant energy. Keeping a camera running continuously on a device as small as an AirPod would decimate battery life. A voice-triggered activation model sidesteps that problem.

Reports from multiple outlets indicate Apple will include an LED indicator light on each earbud to signal when data is being sent to the cloud. This addresses the obvious privacy concern: how do bystanders know when they are being observed? The answer, at least for now, seems to be a glowing stem.

The Best Camera Is the One You Have

There is a truism among photographers: the best camera is the one you have with you. The same logic applies to AI systems that depend on visual input. A frontier model does not need a 48-megapixel sensor to recognize a book title, read a menu, or identify a plant. It needs enough resolution to process context. That threshold is surprisingly low.

Advertisement

For most situational AI tasks, what matters is presence, not fidelity. A low-resolution infrared camera that is always on your head, ready to snap awake when you ask it a question, may prove more useful than a high-resolution phone camera you have to fish out of your pocket, unlock, and aim. This is the bet Apple appears to be making.

The cameras reportedly will not be used for traditional photo or video capture. They are intended to enhance Visual Intelligence, allowing Siri to see and contextualize the physical environment. It is a narrower use case than many expected, but arguably a more practical one.

Apple has not officially confirmed the product. But with references now appearing in release candidate software and a demonstration video accidentally shipped to developers, a September announcement alongside the iPhone 18 seems increasingly likely.