Meta Vs Google Smart Glasses — Which Company Would You Trust With The Camera?

General
1

I tried reading the privacy pages for Meta and Google smart glasses, but I’m still unclear about what happens to camera footage after it’s captured. I’d mainly wear them while walking my dog around the block and picking up groceries, so the camera could catch neighbors, receipts, and house numbers without me thinking about it.

Which company would you trust more with that camera data, Meta or Google? I’m not sure whether “on-device processing” means nothing gets uploaded, or whether clips can still be stored for account features or reviewed to improve services. Are there particular privacy settings, recording indicators, or data-retention terms I should compare before choosing?

2

The hidden privacy risk is the AI feature, not the shutter. A normal photo may stay on the glasses until you import it, while asking “What am I looking at?” can send an image, audio, location, and prompt to company servers. “On-device processing” only describes a particular operation. It does not mean the whole interaction stays offline.

Right now, I’d lean slightly toward Meta, mostly because its shipping products have clearer, testable behavior: gallery captures remain on the glasses until transferred, and the recording light cannot be switched off without disabling the camera. Google’s upcoming glasses may end up better, but I wouldn’t trust promises made before the final consumer privacy terms and settings are available. Google’s broader Gemini policy allows saved activity to be used for service improvement, with separate retention rules for material seen by human reviewers.

Before buying either, check whether AI activity saving can be turned off, whether camera input is uploaded for visual questions, how long deleted data survives, whether humans can review it, and whether imports or cloud backups happen automatically. I’d require a visible recording indicator that stays active for the entire recording, not merely when a clip is saved. For dog walks and grocery trips, I’d leave visual AI disabled and treat the camera like a phone camera: activate it deliberately rather than letting an assistant continuously interpret the surroundings.

3

Realistically, strangers won’t know whether your glasses are recording, running visual AI, or sitting idle. A tiny indicator may satisfy the product design, but it won’t prevent awkward conversations with neighbors, store employees, or parents at a park. For dog walks, that social friction may matter more than the technical retention policy.

I’m less willing than @xepicloopx to treat clearer current behavior as a reason to trust Meta. Clearer controls are useful, but trust also includes what happens after an app update, whether footage enters your existing cloud-photo setup, and how difficult account deletion becomes. Google has the same problem. Both companies benefit when people feed more material into their AI systems, so neither gets much benefit of the doubt from me.

If choosing today, I’d buy based on the least connected mode: camera works without visual AI, automatic uploads can be disabled, and recordings can be removed locally without syncing first. If that setup isn’t possible, a phone is the less awkward and more obvious camera around the neighborhood.

4

Use the return window as a privacy test: record a blank wall, disable every sync option, delete the clip, then check the companion app, cloud storage, and account export later. I wouldn’t trust Meta or Google based on reputation alone. Whichever makes that deletion process obvious and verifiable gets the edge. If copies remain or the settings are vague, I’d return the glasses.

5

Don’t choose based on which company has the nicer privacy statement. Choose based on where an embarrassing clip could accidentally appear. The bigger everyday risk is often boring account plumbing: automatic photo backup, shared albums, partner sharing, cloud device backups, or a companion app that stays signed in on an unlocked phone.

That makes your existing setup relevant. If your Google account already connects Photos, family sharing, home devices, and years of personal data, adding glasses increases the damage from a bad setting or compromised account. The same applies if your Meta account is tied closely to your social contacts and messaging. I’d rather keep wearable-camera footage away from whichever account is most central to my life. If the glasses cannot be used without connecting that main account, that would weigh heavily against them.

The blank-wall experiment from @openwolf7019 is useful, but I would test ordinary mistakes too. Record a short clip, leave it sitting for a day, and see whether it gets backed up without further action. Check whether deleting it from the glasses removes it from the phone, and whether deleting it from the phone affects any cloud copy. Then look at shared libraries and device backups. A deletion button can work exactly as advertised while another copy remains somewhere you forgot was enabled.

Dog walks create plenty of incidental footage that seems harmless until you watch it later. House numbers, license plates, conversations from open windows, children playing, security keypads, and notifications on your own phone can all end up in frame. None of that requires malicious AI processing. It only requires a clip being synced or shared when you did not expect it.

So my answer is neither company gets automatic trust. I’d pick the glasses that let me keep recordings local, avoid connecting visual features to my primary account, and review every clip before it leaves the device. If both products insist on tightly linking camera use, assistant use, and cloud storage, I would skip both. For walking around the neighborhood, the convenience gain is too small to justify turning my main online account into an archive of everything that happened to cross my face.

6

The hidden downside is that glasses give you a poor idea of what you actually captured. With a phone, you see the frame before recording and can check it immediately. With glasses, you may think you filmed your dog while the clip actually includes a neighbor’s front window, a passing child, or several minutes after you thought recording had stopped.

So I would test the capture process before worrying about which corporate privacy statement sounds better. First, start and stop several recordings without looking at the companion app. Make sure the sound or vibration is unmistakable outdoors. Second, cover the camera and try again to learn whether the glasses clearly report a failed capture. Third, check how quickly you can review and delete the clip. If deletion requires importing everything to the phone first, that is a bad design for casual neighborhood use.

I agree with @openwolf7019 about using the return period, but I would spend more time testing mistakes than intentional recordings. Press the button while wearing gloves. Trigger the assistant near traffic. Put the glasses away while they are recording. A trustworthy camera should make those errors obvious instead of quietly creating a file you discover later.

Between Meta and Google, I would not give either company blanket trust. I would choose the specific pair that has a physical camera control, persistent recording feedback, fast local deletion, and no required visual-AI upload. If neither lets you confirm those points in normal use, the phone wins. The privacy advantage of a phone is not that Apple or Google suddenly becomes harmless. It is that you can see exactly when the camera is open and what is inside the frame.

7

Start by denying the companion app background data, precise location, and photo-library access, then see which features stop working. That exposes the real trust boundary: if basic camera use breaks without broad phone permissions, the product is designed around data collection rather than simple capture.

I wouldn’t trust Meta or Google by default. I’d choose whichever still functions under least-privilege settings, and recheck those permissions after every major app or firmware update.

8

Realistically, almost nobody is going to run the blank-wall test past week one. The deletion checks @openwolf7019 and @not_a_debugger describe are solid, but they assume a level of ongoing effort most people drop the moment the glasses become routine. So the setting that actually matters is the default one out of the box, because that is the config you’ll be living under six months later after you stopped fiddling. Whichever pair ships with the least aggressive defaults, no auto-upload, no visual AI running unless you invoke it, wins by default for lazy users, which is all of us eventually. And I’d watch firmware updates more than the launch policy. A quiet toggle flip during an update can undo every careful choice you made on day one, and you won’t get a notification framed as ‘we turned your uploads back on.’

9

Honestly, the company you trust matters less than which one already owns more of your life. Both make money understanding you. The difference is blast radius. If your whole world runs on Google already, bolting a face camera onto that same login is a worse idea than doing it on an account you barely use. @bytewizard265zone nailed that part, and it’s the thing I’d actually decide on, not the marketing pages.

Where I’d push back a little is all the return-window testing. It’s good advice on paper, but @rootlab is right that nobody keeps doing it, and there’s a deeper problem nobody flagged: ‘stays on the glasses’ still leaves a trail even when the footage never uploads. The AI queries, the timestamps, the location tags, the fact that you were standing outside a specific address at a specific time. That metadata is cheaper to store and more useful to them than your blurry dog clip, and it doesn’t show up in a blank-wall deletion test.

One thing that’s genuinely worth knowing before you buy for neighborhood walks: recording other people isn’t just awkward, in a couple of states it’s a legal question once audio is involved, and glasses record audio by default. So the persistent light @xepicloopx wants isn’t only about being polite, it’s your defense if a neighbor ever gets annoyed. My actual take is simpler than the thread suggests. Buy whichever one you can run on a throwaway account with visual AI off, treat it as a dumb camera, and assume the metadata leaks even when the video doesn’t. If that setup isn’t possible on either, your phone is still the honest option.