Meta added a privacy-safety feature to its AI glasses but is reportedly testing a ‘super-sensing’ prototype
12:00 · July 11, 2026 · RSS APP - AI Security and Privacy
Meta's new feature disables its smart glasses' camera if users tamper with the LED recording light.
Summary
Meta has added a safeguard to its second-generation Ray-Ban Meta smart glasses that disables the camera if the device detects tampering with the recording LED. The glasses already turn off the camera when the LED is covered, and the update extends protection to cases where the light is destroyed or altered. Meta argues that a visible blinking indicator provides a practical warning for bystanders, noting that an audible shutter sound loud enough to be noticed would be unsuitable for wearable form factors.
The company is also acting against attempts to circumvent the indicator, removing related listings from Facebook Marketplace and reserving the option to ban accounts or pursue legal measures against those offering modification services. Recording still requires explicit user action through a button press or the voice command “Hey Meta,” yet critics have questioned the LED’s reliability in daylight or among users unfamiliar with its meaning.
At the same time, the Financial Times reported that Meta is testing a “super-sensing” prototype capable of continuous audio capture and periodic image taking every few seconds. Executives have discussed operating this mode without activating the LED, though raw footage and audio would not be stored or made accessible to users or the company. The prototype would allow later AI queries about encountered scenes or conversations, but it remains in early development and separate from the Aria research glasses shown at Connect.
These moves occur against a backdrop of legal and public scrutiny. A lawsuit filed earlier this year alleges that contractors in Kenya reviewed intimate recordings captured by users’ glasses while training Meta’s models. Photos and videos stay private unless shared with the AI, in which case filtered samples may be reviewed to improve performance. The tension centers on consent: wearers may accept data use for product improvement, while bystanders captured incidentally have no equivalent control.
Why it matters
This article is highly relevant for privacy professionals as it highlights the emerging privacy risks and hardware-level safeguards associated with wearable AI devices. The continuous data collection of 'super-sensing' prototypes poses significant compliance challenges under EU data protection laws, particularly regarding bystander consent.
