Meta testing glasses that record everything users see and hear

Meta's contradictory stance on AI glasses privacy—publicly announcing safeguards while internally testing far more invasive capabilities—reflects a deeper tension within the company. The Tuesday blog post promising to disable cameras when the recording LED is blocked appears designed to address public concern about covert recording, yet the Financial Times report reveals that Meta is simultaneously working on a system that would be vastly more difficult for bystanders to detect. Meta itself acknowledged in a policy paper that an always-on LED light would become "invisible" to observers over time, yet that rationale for keeping the light off in super-sensing mode suggests the company is aware of the privacy-visibility tradeoff and choosing opacity.

The metadata-only storage approach—storing locations, device identifiers, and audio metadata rather than raw images—may technically reduce data retention, but metadata alone can expose sensitive patterns about where a person has been and what they have heard. This distinction matters less when prior incidents show that current glasses already capture explicit moments: the Kenya contractor review revealed footage of bathroom use, sexual activity, and financial transactions. An always-on device operating throughout the day would inevitably amplify this risk. The key unresolved question is whether Meta can make the prototype work at all, since continuous recording would drain battery power rapidly. If the company cannot solve the engineering problem, the privacy debate remains academic; if it can, the glasses would represent a significant shift toward ambient surveillance devices.