Sam Altman does not like smart glasses. Speaking on the Sources Podcast, the OpenAI CEO said he finds it “very uncomfortable talking to people with a camera and a light” and confirmed that he does not wear smart glasses himself. That is a surprisingly firm position considering where the technology industry appears to be heading and the success of products like Meta’s Ray-Ban glasses.
Then again, Altman is working with former Apple design chief Jony Ive on an entire family of AI devices. OpenAI has kept the Ive project deliberately foggy, but Altman has suggested they are making a range of AI hardware. His comments about smart glasses draw one clear boundary, removing the idea of a camera pointed at everyone as a feature.
Smart glasses have visible drawback
Altman does identify a fundamental flaw in smart glasses that has little to do with processors, battery life or AI model choices. Glasses place technology directly between two people looking at each other, and adding a camera turns an ordinary conversation into one where the other person has to wonder whether they are also an unwilling piece of content.
Smartphones established a useful social signal around recording. Someone pointing a phone camera at you is usually fairly conspicuous. Camera glasses erase much of that distinction because their great design achievement is making a recording device look inconspicuous. It’s why similar tools have been a staple of James Bond-style spy fiction for decades.
Australia’s eSafety commissioner has already asked manufacturers to make recording indicators impossible to disable and automatically blur faces when informed consent is absent. The regulator cited women being covertly filmed for social content and rejected the idea that telling customers to behave responsibly amounts to a safety policy. The backlash reflects a public tired of paying the social price for somebody else’s convenience.
Important exceptions to the argument against camera glasses exist. The same visual awareness that creates privacy problems can be genuinely transformative for blind and low-vision users, something Australia’s regulator explicitly acknowledged. That makes an outright rejection of the technology too simplistic, but it also strengthens the case that camera glasses may be better suited to particular needs rather than as the default AI hardware.
Wearable, not watchable
Altman’s comments about AI hardware point to an all-encompassing strategy of making devices for tables, pockets, and “on your body,” and that OpenAI will take time to launch all of them. OpenAI has previously described its work with Ive as an attempt to move beyond traditional interfaces. So not just a miniature smartphone strapped to your face.
An audio-first wearable suddenly makes far more sense. An earpiece, pendant, clip or other discreet device could provide persistent access to an AI assistant without placing a camera directly in the sightline of everyone the wearer encounters. It could listen when summoned and respond quietly like wireless earbuds or smartwatches.
Some may prefer the camera and the data an AI assistant gains from it about the physical world. And of course, putting microphones on wearables raises privacy questions just fine without cameras. Making the device unobtrusive yet observant is the trick Ive and Altman have to perform lest it face the same social opprobrium as the AI smart glasses.
Ive built his reputation around products whose complexity receded behind their physical design. But AI assistants will know and hear far more than an iPod ever did. A successful AI wearable cannot merely make advanced computing invisible to its owner. It must make its intentions legible to everyone else.
Meta is betting that the camera can become socially acceptable because they are built into discreet smart glasses; OpenAI may instead bet on a successful AI wearable that won’t potentially creep out everyone you pass on the street. The best OpenAI wearable feature may in fact be an absence, and “no camera” its most enticing selling point.
