"I need your clothes, your boots, and your motorcycle."
Joseph Cox
In response to my posts about Meta's Ray-Ban smart glasses (and "Orion" AR glasses), a number of people pointed to this story in response. The implication here, of course, is that what some see as cool products or technology have downsides – which is, of course, true of basically every technology. Or that "Meta gonna Meta" – of course, this isn't an officially sanctioned use of their glasses and is, in fact, against their terms of service. Still, it's worth talking through this a bit:
A pair of students at Harvard have built what big tech companies refused to release publicly due to the overwhelming risks and danger involved: smart glasses with facial recognition technology that automatically looks up someone’s face and identifies them. The students have gone a step further too. Their customized glasses also pull other information about their subject from around the web, including their home address, phone number, and family members.
The project is designed to raise awareness of what is possible with this technology, and the pair are not releasing their code, AnhPhu Nguyen, one of the creators, told 404 Media. But the experiment, tested in some cases on unsuspecting people in the real world according to a demo video, still shows the razor thin line between a world in which people can move around with relative anonymity, to one where your identity and personal information can be pulled up in an instant by strangers.
What's perhaps most interesting here is that none of this technology – or data availability, for that matter – is new, but it's how the technology is being implemented here which is shocking to people. Said another way: you could already do everything these students did without Meta's glasses. You could do this on your computer. Or even easier, on your phone. But there are still some societal norms that make shoving your phone (let alone a laptop) in someone's face to take a picture of them, not cool. Of course, you could also take a picture with your phone in a more "stealth" manner and still use this technology. But with smart glasses, the cameras are always pointed up at people. And while they have lights to indicate action, we're all going to grow numb to those, fast. And bad actors can just remove those or tape over them, of course.
So again, this is really just about technology continuing to remove the friction to all kinds of things. In many cases, this is great – which is in part why Meta's Ray-Bans are successful: listening to music without earbuds through sunglasses you're already wearing, for example – but again, there are downsides. At the end of the day, bad actors are going to act badly,1 and we're always giving them new tools to do so, sadly.
The demo video posted to X on Tuesday shows the pair using the tech against various people. In one of the first examples, Ardayfio walks towards the wearer. “To use it, you just put the glasses on, and then as you walk by people, the glasses will detect when somebody’s face is in frame,” the video says. “After a few seconds, their personal information pops up on your phone.”
In another example, the demo shows a test on what it describes as “a REAL person in the subway.” Ardayfio looks at the results of a face match on his phone, and then approaches a woman he calls Betsy. He introduces himself and claims the pair met through a particular foundation, presumably referencing something included in the search results.
Yes, this is full-on Terminator mode – aside: who would have thought that for all the negative AGI connotations of Terminator, this facial recognition technology would draw more outrage! – and yes, as creepy as it is, it could be useful in certain circumstances. Highlight – an old school location service to tell you when you're nearby someone you know – eat your heart out.
“The motivation for this was mainly because we thought it was interesting, it was cool,” Nguyen said. When the pair started to show their project to others, “a lot of people reacted that, oh, this is obviously really cool, we can use this for networking, I can use this to play pranks on my friends, make funny videos,” Nguyen said. Then, some mentioned the potential for stalking. Nguyen gave the example of “Some dude could just find some girl’s home address on the train and just follow them home.”
Ardayfio told 404 Media that when the pair did show the technology to other Harvard students and people on the subway, some said, “Dude, holy shit, this is the craziest thing I’ve ever seen. How do you know my mom’s phone number?”
I don't know the solution here. But I do know there's no putting such genies back in their bottles. Meta is not going to stop selling their Ray-Ban glasses, but even if they did, someone else would release them. And governments aren't going to ban such glasses – well, some might, but not Western governments – so here we are. Beyond making such use cases against terms of services, one could imagine companies trying to block the use of such software on their products. But that would require pretty strict lock-down/oversight of app stores – and that's the very thing the EU and elsewhere are currently at war with Apple over!
It's almost like this is more complicated and nuanced than it may at first appear and that there are trade-offs! Funny that.
Speaking of Apple, a few of the responses pointing to this story were doing so with the notion that while Meta may be perceived to be "ahead" in the wearable computing space now, it's in part because Apple would never allow for the creation of technology that would enable such nightmares. That is, of course, ridiculous. You could do all of this on the Vision Pro right now,2 but presumably you wouldn't want to be caught dead wearing that in public. And if you did, people would probably run you out of the subway in the middle of your Terminator doxxing experiment. But really, all of this would probably be not that much harder to do with an iPhone – and certainly it would be faster.
So let's unclench those particular pearls just a bit.
1 In a way, I'm reminded of the other debate currently going on around generative AI. Because you can generate anything, you can generate bad things. That's true. You could also always draw such things. But this technology makes it too easy and good! Sure, that's a problem, but it's not one that is just going to go away. That's a bad actor problem.
2 As mentioned above, presumably Apple wouldn't allow an app to do such things on the App Store -- but it's early enough that perhaps they wouldn't even realize what is was meant for. Also, you can probably do at least some of this through the web browser? At the very least, anyone with intent could just hack any device to do this.