They question the privacy of Meta’s glasses

An investigation has revealed problems in the privacy of users who use Meta’s smart glasses, to the point that sensitive videos, in which they appear naked or in the bathroom, end up being viewed by human reviewers in Kenya.

A joint investigation by Swedish newspapers Svenska Dagbladet and Göteborgs-Posten has questioned the privacy of Meta’s smart glasses, which it manufactures with EssilorLuxottica, after speaking with employees of companies subcontracted by the technology firm in Kenya.

These workers, who have remained anonymous to avoid retaliation, are data annotators and train artificial intelligence manually, so that it learns to recognize and interpret what surrounds people.

They also review videos recorded by the glasses themselves. In them, they have seen people in the bathroom, undressing to change clothes, having sex and watching pornography. They have entered people’s homes and their daily lives, where they have even been able to see their bank card numbers.

Although it has been published recently, the investigation spans several months and is part of the launch of the glasses in Sweden, which allows journalists to go to stores in the country where they are sold. In them, the sellers explain that they are safe to use, although they do not know what happens to the users’ data or where they go.

The use of artificial intelligence on these devices requires an internet connection, because it cannot be executed locally. In their tests, the journalists detected that the data traffic connected to Meta servers in Sweden and Denmark.

When asking the company, a Meta spokesperson ended up referring to the terms of use and privacy policy. “When live AI is used, we process that media in accordance with Meta AI’s Terms of Service and Privacy Policy,” it said in a brief statement.

These policies initially indicate that the user is in control of their data, but they also clarify that voice recordings – necessary to interact with the assistant – can be saved and used to improve AI or other Meta products. Something similar happens with video recording.

Likewise, Meta warns that it can review the interactions that users have with the AI, either automatically or manually, and even warns the user that they should not share sensitive information with the AI ​​that they do not want to share.

Regarding the sending of data to countries outside the European Union such as Kenya, an unidentified Meta manager has indicated that although user data is protected by the General Data Protection Regulation, in reality it does not matter where the servers or companies are located, but rather that they comply with their requirements.