Meta is in the process of testing a prototype pair of “super sensing” AI glasses that use cameras and microphones to constantly record wearer’s actions, the Financial Times has reported.

Citing people familiar with the matter, the paper said that Mark Zuckerberg’s company is developing an updated version of its smart glasses that would continuously record audio while taking photos every few seconds. The wearer could then ask Meta’s AI to help them recall their day or ask follow-up questions about what they have seen or heard.

These glasses have reportedly led to internal debates over privacy challenges, including concerns from non-users about non-consensual recording.

Meta’s current range of AI smart glasses have an LED on their frames that lights up when a user is taking photographs or videos. However, the FT reported, executives do not currently plan to activate the LED when using AI features, making it harder for bystanders to know when they are being recorded. People familiar with the matter told the paper that this is subject to change.

They added that these features could be activated on Meta’s current generation of glasses though a software update.

The $1.5 trillion social media giant is considering several potential workarounds to the privacy issues, including not allowing users or Meta to access the raw footage and audio captured by the glasses. Instead, the metadata from these recordings would be extracted and uploaded to Meta servers to be queried by its AI systems. Proponents reportedly argue that this has fewer privacy implications.

The tech giant is also discussing internally whether data collected through the glasses could be used to train its own AI models, which Zuckerberg said on Monday have progressed more slowly than expected.

Meta declined to comment to the FT on “internal prototypes” but said its approach focused on “privacy built from the ground up”.

---