Source – timesnownews.com
Meta’s Data Usage through Smart Glasses
Meta has confirmed that any image shared with its AI via Ray-Ban Meta glasses can be used to train its artificial intelligence models. This information, initially unclear, was later clarified when Meta informed TechCrunch that photos and videos shared with Meta AI in regions where its multimodal AI is available, like the U.S. and Canada, could contribute to enhancing AI systems. A company spokesperson emphasized that images or videos captured by the Ray-Ban Meta glasses are not used for training unless the user submits them to AI for analysis. However, once analyzed by Meta AI, these images fall under different policies, enabling their use in AI training.
Meta’s first consumer AI-powered device is a step towards collecting vast amounts of data to enhance future AI models. Users who do not want their data to be utilized for training are advised to refrain from using the multimodal AI features offered by Meta. While Meta asserts that its user interface makes this clear, the potential for users to unknowingly provide images of personal spaces, loved ones, and sensitive data is a concern. This move follows the company’s previous approach of training AI models on public data from platforms like Instagram and Facebook.
Recent Feature Launch and Its Implications
Meta recently rolled out new features that simplify how Ray-Ban Meta users interact with Meta AI, further increasing the likelihood of users submitting data to the AI. This comes after Meta’s announcement at its 2024 Connect conference of a live video analysis feature, allowing users to send a continuous stream of images from their smart glasses to the AI system. This feature, promoted for helping users with tasks like picking an outfit from their closet, also enables Meta to use those images for AI training. Although Meta’s privacy policy outlines that interactions with AI features may be used to train models, the company has been reluctant to provide more detailed clarifications on these practices.
In response to queries, Meta’s spokesperson referred to the AI terms of service, which state that by sharing images, users consent to Meta analyzing those images using AI, including facial features. This policy follows Meta’s recent $1.4 billion settlement with the state of Texas regarding its facial recognition practices. In that case, Meta had collected biometric data through a Facebook feature called “Tag Suggestions.” Since then, Meta made facial recognition features explicitly opt-in and deleted previously collected biometric data. However, certain Meta AI features involving image analysis will not be available in Texas, likely due to legal sensitivities.
Wider Industry Impact on Privacy
Meta’s privacy policies also indicate that transcriptions of voice interactions with Ray-Ban Meta glasses are stored to train AI models. Users can opt out of using voice recordings for AI training during the setup process. As Meta and other companies like Snap continue to push smart glasses as a new computing form factor, privacy concerns from the Google Glass era resurface. The potential for privacy breaches is significant with these AI-powered devices worn on users’ faces and equipped with cameras. In fact, some college students have already hacked the Ray-Ban Meta glasses to extract personal details like names, addresses, and phone numbers of people they look at, further raising alarms about the technology’s potential misuse.
Also Read: Meta Unveils Four Exciting Upgrades for Ray-Ban Smart Glasses
