I tried Google’s AI glasses and Meta’s finally got real competition | The Verge

Posts from this topic will be added to your daily email digest and your homepage feed.
Posts from this topic will be added to your daily email digest and your homepage feed.
Posts from this topic will be added to your daily email digest and your homepage feed.
The AI experience is better, the privacy protections are more proactive, and they’re still stylish.
If you buy something from a link, The Verge may earn a commission. See our ethics statement.
Posts from this author will be added to your daily email digest and your homepage feed.
See All by Victoria Song
If you buy something from a link, The Verge may earn a commission. See our ethics statement.
Posts from this author will be added to your daily email digest and your homepage feed.
See All by Victoria Song
It’s a weird time to launch a new pair of smart glasses.
These last few years, Meta has been the dominant player in pushing a fragile smart glasses resurgence. Unfortunately, its privacy reputation led to the “pervert glasses” moniker. Smart glasses have never been more unpopular.
But in the past week, I’ve been demoing some new smart glasses from Samsung, Warby Parker, and Gentle Monster. Some are retro chic frames. Others have the same Y2K vibe as the Matrix films. All were made by Samsung and have Google’s Gemini built in via Android XR, Google’s platform for smart glasses and mixed-reality headsets.
The AI demos are better than I expected. In one, I’m sitting in an avant-garde art installation as creepy heads float above my head. On the loudspeaker, words blare in French, Spanish, Korean, and German. Pressing two fingers to my temple, Gemini starts translating each language seamlessly in real time. In another demo, a TV chef cooks a “monster pasta” recipe in Korean while people have a conversation behind me. Music in another language is playing overhead. This exact scenario tripped up the Meta Ray-Ban Display’s translation capabilities when I was on vacation. Here it’s no big deal. I don’t even have to specify the languages I need translated or, as with Meta’s glasses, download a specific language pair to translate. I sit at another table with an English speaker, a Thai speaker, and the glasses translate everything they say to me — correctly — into Japanese.
But I was most surprised that the glasses automatically redact faces from any photo or video to protect bystander privacy. I asked Gemini to identify my video director, Owen Grove. Gemini told me it knew I was looking at a person, but it can’t identify him because of privacy protections. We pull up a picture of our editor-in-chief, Nilay Patel. There, too, Gemini said that for privacy reasons, it can’t identify a person from a photo either.
Meta’s glasses also can’t do this. But what’s striking here is that Google is making an active choice to take facial recognition completely off the table.
The Warby Parker and Gentle Monster glasses have the customary privacy LED light that lights up whenever there’s an intent to save visual data. Like Meta’s glasses, there’ll be protections in place to disable the glasses if tampering is detected. However, when I sit down to chat with Juston Payne, Google’s senior director of product management for XR, he says Google intends to go much further than the LED light.
“Somebody might come up with some solution [to tamper with privacy LEDs] that our algorithms can’t detect, so then the question becomes: So what can we do?” says Payne.
“We’re not going to target ads on it. We’re not going to train Gemini on it. Humans aren’t going to be able to use it to improve the models. We’re also redacting faces before they actually go to the server in the first place, which then gets ahead of certain things that can obviously go sideways if you don’t take those steps proactively.”
It’s not just faces that’ll get redacted either. Payne tells me that ambient noise — like bystander conversations — will also get scrubbed. The glasses will only process audio from the glasses-wearer using beam-forming mics. Camera and mic access will also require opt-in permissions.
Alone, these are a bold departure from Meta’s approach to smart glasses. Locally redacting faces and ambient audio, for example, is a conscious choice to give up a valuable source of training data. Google could leave it at that, and Android XR would already have a leg up in the privacy discussion. But Payne says Google plans on using its entire platform to disincentivize glassholes.
For example, YouTube will crack down on any creepy videos featuring harassment or unwanted sexualization, regardless of how they’re captured. Google Shopping and Google Ads won’t allow sales or ads for accessories like lens covers for privacy LED lights. Google Play policies will prohibit third-party apps from enabling social livestreaming, real-time facial identification, or changing anything about how the LED light functions. Payne also says third-party apps requiring camera or mic API access will need further review to prove they’re being used responsibly. This is less of a concern right now, but it’s a future-focused stance for other Android XR devices or even with devices like Xreal’s Aura glasses.
It’s hard to gauge how effective this privacy policy will be with the public. We’ll have to see how it’s received. It’s much more expansive than Meta’s more reactive approach so far. Granted, Meta doesn’t have its own app store or a computing platform quite like Android. This still underscores how much more Meta could’ve been doing all along.
The one area Meta truly nailed with smart glasses has been style. The company, through its partnership with EssilorLuxottica, has over 100 combinations of frames and lenses. Samsung’s partnerships with Warby Parker and Gentle Monster aren’t too shabby to look at. Warby Parker has three designs and three colors: green, brown, and black. Gentle Monster has six frames — including two that are very Y2K — with bolder designs, colors, and tinted lens options. The Warby Parker collection