Meta Will Let Smart Glasses Users Opt Out of AI Training on Visual Data

Meta says users of its smart glasses can now opt out of having their "visual data" used to train the company's AI models. The concession, announced as the company battles an ongoing privacy backlash over its AI glasses, also blocks opted-out images from being shown to third-party contractors in other countries.

The change targets a data pipeline that most glasses owners likely never knew existed. Meta has long claimed that photos captured with the glasses' cameras are not shared or used for training. But images and video captured during interactions with Meta AI — for instance, asking "Hey Meta, what does this sign mean" — have been eligible for exactly that purpose. Because these captures are not saved to the user's camera roll, many people presumably do not realize the images are stored at all, let alone used to improve Meta's models. The company has done a poor job of explaining this distinction to users, according to the source report.

Meta framed the new setting in concrete terms. "If you've opted out and you ask your glasses to translate a menu, we'll use your visual data to answer your question and then it's gone," the company explained. "It isn't used for training and nobody reviews it for product improvement."

The stakes behind the policy shift go beyond user confusion. Recordings captured through Meta AI have sometimes been sent to third-party contractors in other countries, who review and label the images for AI training. A report earlier this year found that at least some of those contractors encountered highly personal and intimate images of people while doing this work. That revelation fueled the privacy criticism now pressuring Meta, and at least one proposed class-action lawsuit has targeted the practice.

The opt-out has limits. It covers only visual data. Audio interactions with Meta AI remain available to the company for training and human review, though Meta does let users manually delete those recordings. The setting therefore addresses the most sensitive part of the pipeline — camera captures reviewed by offshore contractors — while leaving a substantial share of user interactions in the training pool.

The timing is notable. Meta has just revealed a new lineup of Ray-Ban-branded glasses that ship without cameras at all, a product decision that signals the company sees camera-free wearables as a way to keep buyers who are uneasy about always-on lenses. Together, the opt-out and the camera-free lineup show Meta trying to defuse privacy concerns without abandoning the AI training practices at the center of its wearables strategy.

For now, users who want their visual data excluded must take action themselves: the opt-out is not the default. That design choice means the burden of privacy protection still falls on the user, even as Meta positions itself as responsive to its critics — and as regulators and courts increasingly scrutinize what happens to the images people's glasses quietly capture.