Meta’s Ray-Ban smart glasses raise privacy concerns over AI features

Meta has confirmed that its Ray-Ban Meta smart glasses may utilise visual and audio inputs provided by users to enhance its smart assistant capabilities. This development, which has been acknowledged by Meta through its policy communications manager Emil Vazquez, involves using images and videos shared with Meta AI for training purposes in alignment with the company’s Privacy Policy.

The smart glasses feature a tool known as “Look and Ask”, where users can take a photo of a subject to make queries such as requesting information about a landmark or having a sign translated. The AI analysis for these functions relies on cloud processing, requiring users to share their images with Meta. This means that while using these specific AI features, users are allowing their submissions to be used in training Meta’s AI systems.

Currently, these functionalities are limited to users residing in the United States and Canada. People in regions without access to Meta AI or those who bypass the AI analysis tools of the smart glasses can rest assured their images remain private, barring any uploads to platforms like Facebook or Instagram where Meta’s policies might permit AI training using those posts.

While some may find this utilisation of personal photos unsettling, it’s consistent with practices by other AI developers who harness user inputs for developmental purposes. In contrast to on-device AI technologies promoted by companies like Google and Apple, which tout enhanced privacy, the cloud-based nature of Meta’s assistant necessitates data sharing for operation.

For users conscious of privacy, it’s important to note that opting into the AI analysis feature means giving consent to share images, with the only current recourse being to discontinue use of the AI features to avoid participation. Meta’s method diverges from systems where the AI interaction is not constantly present; wearing smart glasses with such capabilities often entails a persistent interaction with the device.

Moreover, as voice-activated AI has become more intuitive and accessible, there is a heightened risk of involuntary sharing if users aren’t fully vigilant. The ease of activating Meta’s assistant through natural language inputs could lead to unintentional data submission.

Meta is yet to disclose any future strategies to better inform users about data usage or to introduce more nuanced opt-out configurations that keep user functionality intact. In the interim, users of Ray-Ban Meta smart glasses might need to stay alert when choosing to engage with the AI-enabled tools, considering the broadest implications of sharing data in this manner.

Source: Noah Wire Services