Meta Finally Lets Smart Glasses Users Opt Out of AI Training

Meta has started rolling out a setting that lets smart glasses users opt out of having the company use "visual data" collected by the device for AI training. The setting is now live inside the Meta AI app, and the company has begun notifying glasses owners with a pop-up.

The notification spells out what is at stake. "Your AI glasses can translate a menu, identify a plant and more," it reads. "You can opt out of providing visual data from these experiences to Meta for product improvement. If you opt out, that data isn't stored after processing."

The pop-up directs users to the relevant settings page, but the setting can also be found manually. Users need to navigate to the settings for their glasses, select "Meta AI" and scroll down to "store visual data from AI experiences." The glasses must be out of their case and connected to a phone to change the setting. Anyone with multiple pairs of glasses must repeat the process for each pair.

The opt-out comes with a cost. Meta says turning it off may prevent users from accessing certain features, including "visual reminders" — looking at an object and asking Meta AI to remind them to buy it next week. It is not clear whether any other features are affected.

Alongside the new setting, Meta has published a more detailed description of how it uses visual data in a help center article. The article explicitly acknowledges the possibility of human review. "When you use AI experiences on your Meta glasses – like asking about something you're looking at or using live AI – Meta stores the resulting visual data and may use it to power glasses features and improve Meta products," the company explains. "This includes the process of training and improving AI models, which may be automated or manual (human)."

In other words, photos and video captured during AI interactions can end up in front of human contractors, not just in an automated training pipeline. The new setting determines whether that data is stored at all after processing.

Meta notes that it only stores visual data collected while using certain AI features, including Live AI. The company defines visual data narrowly: it does not include "photos and videos you capture with your glasses when the capture LED is on." Regular photos and videos, taken with the recording indicator lit, sit outside the scope of this data collection.

The distinction matters. Meta's glasses-based AI features — such as continuously running live assistance — process whatever the wearer is looking at, often without an explicit capture event. That continuous visual stream is the data Meta stores, uses to improve its products, and may submit to human review unless the user opts out.

For buyers of Meta's glasses, the practical takeaway is simple: the opt-out exists, but it is not on by default, and finding it requires several taps through device settings with the glasses connected. Users who skip the pop-up will continue to share visual data from AI interactions with Meta by default.

The rollout signals how Meta is handling the tension between its AI training ambitions and the privacy expectations of people wearing camera-equipped computers on their faces. Smart glasses with always-available AI assistance sit in a gray zone between a camera and an assistant — and regulators and users are still deciding what consent should look like there.

The new setting also gives Meta something concrete to point to as scrutiny of AI training data practices intensifies across the industry. Whether an opt-out buried in device settings satisfies that scrutiny — or simply becomes the template for how wearable AI products handle visual data going forward — is the question the company has now, deliberately or not, put in front of its users.