Each earbud takes its own image, which is then synchronized for use by Visual Intelligence.

Apple accidentally leaked a video of its forthcoming camera-equipped AirPods earlier this week, thanks to a test version of an upcoming macOS update. Now we have even more information about the unique earbuds, also thanks to code buried in that macOS Tahoe 26.7 early build.

The idea of camera-equipped earbuds is pretty wild on its own, but the details on how this will likely work are downright weird. Each earbud will reportedly have its own camera and images will be synchronized by the Visual Intelligence software into a cohesive whole. The cameras will take RGB still images up to just 1 megapixel.

The AirPods will account for head movement, which isn't a surprise as the company's earbuds already have multiple accelerometers. However, there's some type of software that will reject images impacted by excessive movement or by an occluded view. Some of this processing looks to happen on the actual earbuds, which is impressive, as the code mentions "peripheral inference." This sounds a whole lot like on-device detection.

It's also worth noting that the AirPods do not shoot video. Instead, they support periodic image capture. This will reportedly be enough for the Visual Intelligence software to understand.

There's both an active and passive mode. This looks to change up the resolution a bit, with the active mode producing higher-quality images. It's also likely that each earbud includes a capture indicator light, so people should know when the cameras are being used.

The capture indicator light is always good from a privacy perspective but these are not Meta Glasses. They aren't really intended to take images of the world for personal use. They are intended to take images of the world for use by Apple's forthcoming Visual Intelligence software.

This is a part of Apple's long-delayed upgraded Siri AI assistant. The multimodal assistant will reportedly be able to analyze the images sent by the earbuds and, well, do stuff with them.

The only actual use case scenario described by Apple, in a short leaked video, shows a man asking Siri to remember a book he's looking at. A voiceover states "with Visual Intelligence, your world becomes saveable. See something you like? Just ask me to save it for later." The human brain also does that. It's a core feature called memory.

A major question mark here is how accurate Visual Intelligence will be when accessed by the earbuds, given the low-powered cameras. AI models are already prone to mistakes, even when processing pristine data.

We might not have all that long to wait for answers. Apple is holding its big iPhone 18 event next month and we could get more information at that point, given the demo video's presence in macOS 26.7. However, Bloomberg's Mark Gurman has suggested they won't be released until 2027.