HTC's New Smart Glasses Offer a Few Imperfect Fixes for the Pervert Problem

Privacy and smart glasses should go hand in hand, and with the $499 Vive Eagle AI , HTC seems to realize that. It's clear the company wants to avoid the "pervert glasses" stigma that Meta's smart glasses have earned, and it's taking notable steps toward that end, though the built-in camera remains on the Vive Eagle AI. I'm currently testing them, and a full review is coming, but for now, here's what I see with HTC's "privacy-first approach."

Addressing and Anonymizing AI

Like all AI glasses, the Vive Eagle has built-in voice-activated AI. Interface functions, like activating different features, are handled by HTC's Vive AI, and HTC says your data will not be used to train its AI. Frankly, I think all AI data training should be opt-in, not the opt-out-and-good-luck-finding-out-how-because-we-won't-tell-you approach of a lot of AI systems. Including AI systems you weren't interacting with in the first place , like Twitch .

If you want general information beyond the glasses' core functions, the Vive Eagle supports Google Gemini and OpenAI's GPT. They're much larger systems that have basically scraped the entire internet for training data, and will certainly use your interactions as well.

Considering how much Google is tied to our internet footprints, there's a reasonable concern that every Gemini-related thing you do with the glasses will be linked to your Google profile. However, HTC says LLM requests through the glasses will be de-identified. The company pointedly says it makes it "harder" to build a profile, but not impossible. Sadly, that's the current reality we live in, and no privacy or security feature is completely bulletproof. Welcome to the data economy.

The Vive Eagle is also ISO 27001 and ISO 27701 compliant, which means it meets international standards for securing personal data and ensuring some level of user-side privacy control.

The Camera Problem

The Vive Eagle has a built-in 12MP camera that lets you take photos and 3K video, and provides translations in 96 different languages. A camera is always handy to have, but it's also the main reason "pervert glasses" has become a common phrase. Creeps, stalkers, and general jerks use camera-equipped smart glasses to surreptitiously record and harass people. To address the problem, HTC added extra privacy and security features to the Vive Eagle.

First, its recording light uses multiple sensors to ensure it's always visible when the camera is recording. It will adjust its brightness based on ambient light to be easily seen, and the camera won't work if the light is obscured in any way. A recording light that you can't disable has become standard on major smart glasses with cameras, including Meta's , but I'm skeptical that it can't be circumvented in some way.

Second, the camera will not record unless the glasses are being physically worn. This means someone can't take the glasses off and set them to capture from a hidden position. There could also be a workaround there, but it's another good step.

Third, the camera will turn off if anyone nearby says "stop filming." That's kind of brilliant in its simplicity. If you know you're being recorded, you can stop it. Again, in theory.

The fourth feature is less an anti-pervert-glasses fix and more about general user security improvements. Photos and videos taken on the Vive Eagle stay on the glasses' on-board 32GB storage until you manually transfer them to your phone or another device, and that data can't be accessed if you pair the glasses with a different device. When combined with the ISO 27001 and ISO 27701 certifications, anything you capture on the glasses should be safe from other people.

Privacy, Security, and Uncertainty

It's good that HTC is up-front and highlighting its privacy and security measures on the Vive Eagle. They aren't remotely guaranteed, since no information security measure is, but it's a start. Still, in an ideal world, we wouldn't have to worry about this at all.

Security to prevent individuals from stealing your data is one thing. Preventing unauthorized system access by bad actors has been a vital part of information security for as long as information has existed. Thieves will always exist, so we find ways to deal with the threats.

But the concerns about AI glasses go beyond those bad actors to the devices' own manufacturers, service providers, and users. The Vive Eagle doesn't anonymize your AI prompts to keep some random hacker from tracking you; it does that to keep Google and OpenAI from tracking you. HTC comes out and says Vive AI won't use your data for training, even though every major AI firm  is  using it for training, whether you like it or not. The glasses don't have a "stop filming" kill switch to keep your photos and videos out of others' hands; they have it to keep you from being a creep or stalker with them.

These privacy and security features aren't simply necessary to protect us from criminals. They're necessary to protect us from the very services we rely on, and from ourselves (or at least other people using the exact same devices). HTC seems to be doing what it can with the Vive Eagle (besides removing the camera entirely ), but the fact that these actions need to be taken in the first place? Well, that really sucks.

PCMag and Yahoo may earn commission from links in this article.