Meta trains its AI models with images and videos shared with Meta AI through Ray-Ban Meta glasses

Oct. 3 (Portaltic/EP) –

Meta has confirmed that can train their Artificial Intelligence models (AI) with any image or video that users share with the Meta AI assistant through Ray-Ban Meta glasses with the newly introduced new analysis functions.

The company led by Mark Zuckerberg announced last week new AI-powered features and tools coming to your Ray-Ban Meta glasses, within the framework of its annual event Meta Connect 2024, among which it presented the possibility of interacting more naturally with the assistant, as well as the support for video input in Meta AI.

With this last function, users can carry out actions such as exploring a city and have Meta AI accompany the experience by offering information about the landmarks they are visiting in real time. For this, analyzes the images that the user is currently viewing.

Following this line, another of the announced features was that the The glasses’ vision system also allows you to scan visual elements such as text or QR codes and, based on this, carry out actions such as calls.

These options try to provide useful services to users through the Meta AI assistant, however, the company has confirmed that any image that is shared with the wizard to use these features can be used later to train your AI models.

Specifically, the technology company has clarified that these images and videos shared with Meta AI can be used in “places where multimodal AI is available”, i.e. currently in the United States and Canadaas explained by Meta Policy Communications Manager, Emil Vazquez, in statements to TechCrunch.

However, the company has also clarified that does not use photos and videos captured with Ray-Ban Meta glasses as long as they are not used with the Meta AI assistant functions. Therefore, when, for example, the assistant is asked to analyze a video through the glasses to obtain more information about a place being visited, these images can be used to train AI models. .

This implies that the American company can use any image that users are viewing through Ray-Ban Meta glasses to train their new AI models, which may result in share images of personal placessuch as users’ homes, or other types of sensitive information.

However, Meta has indicated that this activity is clearly detailed in the Ray-Ban Meta user interface. Likewise, it has indicated that this use is specified in the Meta AI assistant terms of service, where the company establishes that users have the option to share images with AI and, once they are shared, They agree that Meta will “analyze such images, including facial gestures, with AI.”

For the moment, the Use of this information to develop and improve Meta AI models does not take place in the countries and regions of the European Unionin order to comply with the privacy and security requirements of community legislation, as detailed by the technology company on its website. Privacy Policy.