Apple and Google are exploring a new class of AI hardware that captures audio and video without, in the companies' telling, actually recording it. According to a Bloomberg report cited in a Verge column, Apple is developing a smart home camera that would use AI to generate text descriptions of events instead of saving video footage. Apple's Audio Intelligence features on the latest Apple Watch similarly produce transcripts and summaries of conversations while the company says the raw audio is deleted after processing.
Google's senior director of wearables, Sandeep Waraich, told The Verge that the company is considering smart glasses whose cameras process visual data without saving any footage. The underlying argument is that a recording only exists if something is preserved and reviewable. But as The Verge's Victoria Song points out, the AI-generated text snippets and transcripts are themselves preserved and reviewable, which undercuts the semantic distinction.
The stakes are commercial as well as legal. Redefining 'recording' could make always-listening devices more palatable to consumers and regulators, but it also complicates consent. Meta's smart glasses signal recording with a white light, not the universal red, and other AI wearables may have no indicator at all. Song argues that if cameras and microphones are everywhere but 'not recording,' people will have no way to know when they are being captured and summarized.