Meta’s smart glasses are starting to take off, with the company claiming it has seen “unprecedented demand” for its AI-powered Ray-Ban shades. But users are seemingly unaware that Meta sends the recorded footage to data annotators whose job is to review and label the content as a way of training AI.

These employees are seeing some things they wish they weren’t, which brings privacy into question.

“AI smart glasses raise significant privacy concerns,” Kleanthi Sardeli, data protection lawyer at nonprofit None of Your Business, previously told Reuters. “The main issues are linked to the use of people’s personal data to train AI models and transparency for bystanders.” Meta’s Data Annotators Are Viewing Private Content

“In some videos, you can see someone going to the toilet or getting undressed,” a contractor for company Sama told Swedish newspapers Svenska Dagbladet and Göteborgs-Posten. “I don’t think they know, because if they knew they wouldn’t be recording.”

Featured Video

How Canva Became the Power Player in the AI Design Wars

“I saw a video where a man puts the glasses on the bedside table and leaves the room,” another data annotator said. “Shortly afterwards, his wife comes in and changes her clothes.”

It’s uncomfortable for the employees, who feel they don’t have a choice.

“You understand that it is someone’s private life you are looking at, but at the same time you are just expected to carry out the work,” one of them said. “You are not supposed to question it. If you start asking questions, you are gone.” The Fine Print

The review process is disclosed in Meta’s AI terms of use, where the company says it has the right to “review your interactions with AIs, including the content of your conversations with or messages to AIs, and this review can be automated or manual (human).”

It continues that users shouldn’t share information that “you don’t want the AIs to use and retain, such as information about sensitive topics.”

But it’s clear from the data annotators’ work that people aren’t grasping this warning.

All users have to do to start recording is press and hold a button on the frame of the glasses. It’s easy to imagine a user forgetting to hit stop, leaving the glasses to record whatever they do next. The problem is that as soon as the footage has been captured, it isn’t theirs anymore.

“Once the material has been fed into the models, the user in practice loses control over how it is used,” lawyer Sardeli told the newspapers.

“When live AI is being used, we process that media according to the Meta AI Terms of Service and Privacy Policy,” a Meta spokesperson said in a statement to the outlets.