Are Google AI glasses private? What Gemini keeps from the camera and from your words
The Morse teamUpdated 10 min read
As of November 2026, Google's AI glasses are private about what the camera shows Gemini, but not about what you say to it. Google, which opened preorders for its Gemini glasses on November 12 (Google), says the camera images Gemini uses aren't saved in your Gemini History, aren't used to train AI or target ads, and have bystanders' faces redacted first, while your chats with Gemini are saved in Gemini History under settings you control (Google). Those chats fall under the same history and training rules as Gemini on your phone.
We read Google's two November 12, 2026 launch posts, the glasses product page and its FAQ, the Google Store page, the Google Safety Center page for the glasses, the Gemini Apps Privacy Hub and Privacy Notice, Google's Gemini activity help page, and Android XR's app policies for glasses, all on November 14, 2026, and re-read the Privacy Hub and the history help page on November 17, 2026. Google earns credit here. It put the bystander protections into the hardware, with a light that, when covered, makes the glasses stop recording and refuse to take photos, and it wrote rules for third-party camera apps before the glasses shipped.
What does Google keep from its AI glasses?
Google sells the glasses as "intelligent eyewear," made with Samsung, Warby Parker, and Gentle Monster. Warby Parker's frames start at $395 and Gentle Monster's at $449, they go on sale December 10, and they work only with 18+ Google Accounts from the US and South Korea (Google). These are the rules on Google's own pages, read November 14, 2026.
| What | What Google says | Source |
|---|---|---|
| Camera images Gemini uses to answer you | Not saved in Gemini History, not used to train AI models or target ads | |
| Bystanders' faces in those images | Redacted on the device before Gemini processes them | Google Safety Center |
| Your chats with Gemini | Saved in Gemini History; a setting controls whether their text trains AI models | |
| Photos and videos you take | Sent straight to your phone; the outer light shows while you capture | Google Safety Center |
| Covering or tampering with the light | The glasses are designed to stop recording and refuse to take photos | Google Safety Center |
| Microphones before "Hey Gemini" | Standby; nothing sent to Google until an activation, which a similar-sounding noise can trigger | |
| Data on the glasses and in transit | Encrypted | |
| Third-party apps | Must ask before using the camera; you can revoke access per app | |
| Power switch | Turns off the camera, microphones, and Gemini | Google Safety Center |
| Private AI Compute | "Coming soon," no date given |
Do Google AI glasses record you?
Only when the wearer takes a photo or video, according to Google. The Safety Center says the glasses "are designed to only take photos or videos for your gallery when you tell them to," by pressing the camera button or asking Gemini (Google Safety Center). Those files go "straight to your phone," and you choose whether to back them up to Google Photos or another app.
The microphones wait for a wake word. Google's FAQ says Gemini on the glasses stays in standby until it hears "Hey Gemini," and "When in standby mode, Gemini won't send what you are saying to Google or anyone else" (Google). The same answer adds that a noise that sounds like "Hey Gemini" can wake it too. You can also start Gemini with a tap on the frames (Google). Asking Gemini about your surroundings needs a Bluetooth connection to your phone, while taking a photo, checking the battery, and changing the volume don't (Google).
Gemini tries to hear only you. Google says it designed Gemini "to focus on your voice by default, and not other voices around you, using directional microphones on the glasses" (Google).
Does Google save what the glasses' camera shows Gemini?
No. Google's privacy post says "When you ask Gemini on glasses about the world around you, like getting help translating a menu in another language, the camera visuals are not saved in your Gemini History," and the visuals "are not used to train AI models or target ads" (Google). A footnote sets one limit: "If you ask Gemini to take a photo or video, it will store it in your gallery."
That rule is stricter than Gemini Live on a phone. Google's Gemini activity page says a Live chat's activity can include "video frames of your camera stream" when Gemini History is on (Google), as we cover in is Gemini Live private. The glasses' Guided Vision runs "in Gemini Live" (Google), so the glasses carve their camera out of Live's usual rule.
The carve-out isn't in Gemini's privacy notice. On November 17, 2026, the Gemini Apps Privacy Hub, last updated November 16, 2026, didn't mention glasses at all, though it lists "Gemini on Android XR" among its apps (Google), and the activity page that the glasses FAQ links to still describes only phone and web Gemini. The no-save promise appears in Google's launch posts, its product page, and the Safety Center, which are Google's own words but not its privacy notice.
Does Google train AI on what you say to the glasses?
Not by default, by Google's newest help page. As of November 2026, it says the settings that let Google train generative AI on your chats, Chat content including text & uploads and the audio and Live one, "are off by default" and need History on (Google). The glasses need an 18+ account, and for adults "History is on by default." Google's page doesn't say whether accounts that had Keep Activity on before were switched over, so check Settings, then History. Google's privacy post says "Your chats are saved in Gemini History," that you "control in settings whether the text of your chats is used to improve generative AI models," and points to the Gemini Apps Privacy Hub for the rest (Google). "Gemini History" is also what Google's help pages now call it, controlled by the History setting at gemini.google.com under Settings, then History (Google), and none of the glasses pages we read gives a path to it in the glasses app. The defaults, retention, and reviewer rules are the ones we cover in is Gemini private and is Gemini Live private.
Do Google's glasses blur bystanders' faces?
Yes, in what Gemini sees. The Safety Center says "When you ask Gemini about what you're looking at, your glasses are designed to automatically redact bystanders' faces before responding," under a heading that calls it an "on-device" protection (Google Safety Center). It also says Gemini on glasses "is designed to not recognize faces of people around you."
Google describes redaction only for Gemini. None of the pages we read says it applies to the photos and videos you save, to third-party apps you give the camera to, or to Aira Explorer, which Google says streams "your glasses' camera feed directly to a professionally trained human visual interpreter" (Google).
Android XR's rules for glasses apps, last updated October 20, 2026, let apps use the camera for video calls and for "ambient camera sensing or scene understanding" (Android Developers). The same rules bar apps from social live streaming, from facial identification that surfaces bystanders' "names, home addresses, phone numbers, or their employer," from "persistent, background audio recording," and from using the microphones "to identify non-wearer voices or bystanders."
Can people tell when Google's glasses are recording?
When you take a photo or video, yes. Google's FAQ says that for a photo "the inner and outer indicator lights flash, accompanied by a photo capture sound," and for a video they "pulse continuously" with a tone when recording starts and stops (Google). The Safety Center says sensors check whether the light is blocked, and "If tape, paint, hair, or anything else covers the light," the glasses are designed to stop recording (Google Safety Center). Hardware checks do the same if the light's wiring is "compromised or modified."
The light is tied to your gallery. The Google Store FAQ says it is "designed to turn on when a photo or video is being saved to your gallery" (Google), and no Google page we read says it lights when Gemini looks at something to answer you. Meta's glasses work the same way: Meta says their light blinks "Whenever content is being captured for your gallery" (Meta), as we cover in are Meta AI glasses private.
Google puts some of the rest on the wearer. Before first use, you agree not to record others without permission, not to record in sensitive places, and not to cover the light, and Google says breaking them is "a violation of the Google Terms of Service" that can get "a user's ability to use the glasses" disabled (Google).
What do Google's glasses protections cover, and what do they leave out?
Google's protections cover what the camera captures for Gemini and for your gallery. As of November 14, 2026, they're spread across several Google pages, and each leaves a gap.
| Part of the glasses | What Google promises | What its pages don't say |
|---|---|---|
| Camera image sent to Gemini | Not saved in Gemini History, not used for training or ads (Google) | The Gemini Apps Privacy Hub, last updated October 26, 2026, doesn't mention glasses (Google) |
| Bystanders' faces | Redacted on the device before Gemini processes the image; Gemini designed not to recognize faces (Google Safety Center) | Whether redaction covers saved photos, third-party apps, or Aira |
| Outer light | Flashes for photos and pulses for video, with a sound; covering it stops recording (Google) | Whether it lights when Gemini looks to answer you (Google) |
| Aira Explorer | Streams the camera feed to a trained human interpreter, on request (Google) | Whether faces are redacted or what Aira keeps |
| Third-party apps | Must ask first; no social live streaming or facial identification of bystanders (Android Developers) | How the rules apply to apps from outside Google Play |
| Gallery photos and videos | Sent straight to your phone (Google Safety Center) | Nothing beyond where you choose to back them up |
For everything else in your Gemini History, Google's privacy post points to the Gemini Apps Privacy Hub (Google), whose rules for voice, location, and reviewed chats we lay out in is Gemini Live private.
Will Private AI Compute make the glasses more private?
Google says it's on the way, without a date. Its privacy post says "Coming soon, we're also bringing Private AI Compute technology to glasses to process data in a secure, fortified space" (Google). The post doesn't say which glasses features will use it or whether it changes what Gemini History keeps. We explain how Google's system works in Google's Private AI Compute.
How Morse keeps your questions out of your Google Account
Google kept the camera out of your history, but every question you ask the glasses still lands in your Gemini History, under your Google Account, where Gemini's training and review rules apply to it. Morse can't change what your glasses send. For the questions you'd rather not say out loud to Gemini, Morse is a web app where you type them, and your chat history is stored in your browser, encrypted, not on Morse's servers (Morse privacy policy).
- Private, for the question you'd otherwise leave in Gemini History. Every new chat opens in this mode. Open-weight models such as DeepSeek, Kimi, and GLM answer under the zero data retention agreements Morse has with its inference, search, and other providers, so the hosts agreed to keep nothing. Where a provider confirms zero retention on each reply, Morse checks it and refuses any reply that fails.
- Anonymous, when you want Gemini's answer without your Google Account attached. Choose Gemini in the model list. Our proxy strips your account, email, and IP address and sends the question under Morse's own account, so the provider sees the message without knowing it's you. It may keep it for a limited time to check for abuse, under its own policy, so describe what you saw without names or addresses.
- E2EE, for what you'd never ask with a microphone on your face. Your browser checks the enclave's hardware attestation before anything leaves your device, then encrypts the message for that enclave alone, and our servers pass along text they can't read. The enclaves are run by third-party providers, not by Morse, and the guarantee rests on the enclave hardware and attestation working as designed. E2EE chats have no web search or image generation, and they come with Pro, at $19 a month.
Morse doesn't train AI on your content, sell your data, or show ads; in Anonymous mode the provider's own terms apply. Its servers don't keep messages or log what you write, and because we don't store chats, we usually have very little to give authorities. Your history locks when you sign out. Clearing your browser or deleting your account deletes your chats for good, and Morse can't recover them, so save anything you'll need later.
So are Google's AI glasses private enough for you?
- The people around you: Google describes face redaction only for what Gemini sees, not for the photos and videos you save. Ask before you film, as Google's own rules require.
- What the camera shows Gemini: Google says it isn't saved or trained on, though as of November 14, 2026 that promise sits outside the Gemini privacy notice.
- What you say: check that Chat content including text & uploads is off under Settings, then History, at gemini.google.com if you don't want your words training Google's models.
- Apps you add: grant camera access only to apps you'd trust with your view, and revoke it per app when you're done.
- Questions you'd rather keep to yourself: don't ask them out loud. Type them in Morse, in Private mode, or in E2EE for anything no one should read.
Try a chat from the Morse home page without an account, or start free with no card needed.
Common questions
Do Google AI glasses have facial recognition?
No. Google's Safety Center says Gemini on glasses "is designed to not recognize faces of people around you," and Android XR's app policies bar apps from running facial identification to surface a bystander's name, address, phone number, or employer.
Can Google turn off my glasses?
Yes. Before first use you agree not to record people without permission, not to record in sensitive places, and not to cover the outer light, and Google says breaking those rules violates its Terms of Service and can get the glasses' functions disabled.
Do Google's AI glasses work without a phone?
Only partly. Google's FAQ says asking Gemini about your surroundings needs a Bluetooth connection to your phone, while taking a photo, checking the battery, and changing the volume don't.
Where can you buy Google's AI glasses, and for how much?
Preorders opened November 12, 2026 at Warby Parker (from $395) and Gentle Monster (from $449), with sales starting December 10. They work only with Google Accounts from the US and South Korea, and the account has to be 18 or over.