Meta adds opt-out to stop AI glasses visual data from training models

Meta announced a new privacy setting for its AI-powered smart glasses that allows owners to prevent visual data from being used to train the firm’s artificial-intelligence models. When the opt-out is active, any image taken to answer a query—such as translating a menu—will be processed and then discarded, and it will not be forwarded to third-party labeling teams abroad. The change does not affect audio recordings, which continue to be stored for training unless users delete them themselves.

The adjustment comes amid mounting privacy criticism and a proposed class-action lawsuit over earlier practices that sent users’ images to contractors who sometimes viewed intimate content. In parallel, Meta unveiled a new line of Ray-Ban-branded glasses that contain no cameras at all. The company maintains that photos saved to the device’s camera roll have never been used for training.