Meta lets smart glasses users say no to AI training
What's the story
Meta is giving users the option to opt out of their "visual data" being used for artificial intelligence (AI) training. The move comes amid growing privacy concerns over its AI glasses. The company clarified that if a user opts out and asks their glasses to translate a menu, the visual data will only be used once and not stored or reviewed for product improvement purposes.
Training concerns
Controversy over visual data usage
Previously, Meta had maintained that the content of photos taken with its glasses' cameras wasn't shared or used for training.
However, images and videos captured during interactions with Meta AI were used for this purpose.
These captures aren't saved to the user's camera roll, leading many to be unaware of their storage and usage.
The company has been criticized for not adequately explaining these practices to users.
Contractor involvement
Data sent to 3rd-party contractors for review
The images and videos captured during Meta AI interactions were sometimes sent to third-party contractors in other countries. These contractors would then "review" and label the images for AI training.
Some of them have even reported seeing personal and intimate images of people while doing this work, raising further privacy concerns about Meta's data handling practices.
Data coverage
Audio interactions still used for training
The new opt-out setting from Meta only covers visual data, not audio interactions with Meta AI.
These audio interactions are still used by the company for training and review purposes. However, users do have the option to manually delete these recordings if they choose to do so.
The move comes as part of Meta's effort to address privacy concerns and at least one proposed class-action lawsuit.