Hardware

Leaked macOS code reveals Apple camera-equipped AirPods

Apple has inadvertently leaked video and code for camera-equipped AirPods in macOS 26.7 RC, signaling a major push toward screen-free, AI-driven visual intelligence.

TechCrunch AI2 days agoHardware
Image: TechCrunch AI

Apple has exposed plans for camera-equipped AirPods within the code of its macOS 26.7 RC software release. Discovered by researcher Aaron Perris, the leak includes a video demonstrating a user wearing the earbuds while holding a book and interacting with Siri. An accompanying audio track explains that the device uses Visual Intelligence to let users save things they see simply by asking Siri. The software also contains a Hair Detected error message to alert users if their hair is blocking the earbud cameras.

The hardware, which social media leaks identify as the AirPods Pro 4, is designed to integrate with the upgraded Siri assistant launching alongside iOS 27 in September. According to earlier reporting by Bloomberg analyst Mark Gurman, these cameras are not designed to capture high-resolution photos or videos. Instead, they act as low-resolution sensors that feed visual data directly to Siri. To address privacy concerns, the earbuds will reportedly feature an LED indicator that illuminates whenever visual data is being transmitted to the cloud.

For AI developers and hardware practitioners, this development represents a significant shift toward ambient, screen-free computing. By embedding visual sensors into a socially accepted form factor like wireless earbuds, Apple is bypassing the social friction that has hindered smart glasses like Meta Ray-Bans, Snap Specs, or Google's AI glasses. Practitioners can prepare for a new class of voice-and-vision applications that rely on real-time environmental context rather than screen-based inputs.

This integration allows Apple to expand its ecosystem without forcing users to constantly look at their iPhones. The low-resolution feed ensures that processing demands and privacy risks remain low, while still providing enough spatial context for tasks like navigation, object identification, or reading assistance. Developers will need to adapt their conversational AI models to handle these continuous, low-bandwidth visual streams.

This is our own summary of reporting by TechCrunch AI

More in Hardware