We recently inquired of Meta whether they train their AI models on the visual content users upload to the Ray-Ban Meta smart glasses, which are designed for augmented reality experiences? .
Since then, Meta has offered TechCrunch additional context, providing more insight into its plans and decisions.
Any image shared with Meta AI can seamlessly train its artificial intelligence.
In areas where multimodal AI is available, users can opt-in to utilize photos and videos shared with Meta AI for enhancement purposes, as outlined in our Privacy Policy.
The company previously stated in an email announcement that images and videos taken with Ray-Ban Meta are used by Meta for training purposes, provided the user does not intentionally submit them to AI processing. Despite this, when you request Meta AI to investigate these images, they are governed by an entirely distinct set of rules.
The corporation is leveraging its inaugural customer-facing AI platform to amass an extensive repository of data, which can be harnessed to craft increasingly sophisticated AI architectures. One approach to avoiding the pitfalls of “deciding out” is simply to refrain from using Meta’s multimodal AI features in the initial stages.
The concerns arise because Ray-Ban Meta users may inadvertently share extensive personal content, including interior views of their homes, family moments, or sensitive documents, which would train Meta’s AI models. Meta’s representatives clarify that clarity exists within the Ray-Ban Meta user interface, whereas the company’s executives initially demonstrated ignorance or reluctance to disclose these details to TechCrunch. We are aware that Meta trains its LLaMA AI models on publicly available text data, which includes a vast amount of web pages, books, and articles. Meta has broadened the scope of “publicly obtainable information” to encompass data people access through their smart glasses or request AI-powered analysis from chatbots.
That is notably related now. Meta recently unveiled new AI features that simplify the process for Ray-Ban users to integrate Meta AI into their daily routines, enabling seamless data sharing and facilitating more effective training. At Meta’s 2024 Join convention last week, the company announced that it will introduce a feature that streams a continuous flow of images directly into its multimodal AI models. Meta touted the technology’s potential to revolutionize personal styling, enabling users to utilize AI-driven insights to curate outfits from their existing wardrobe.
The corporate doesn’t explicitly reveal that by sharing these photos, you’re also contributing them to Meta’s model training processes.
Meta officials redirected us to their statement, which explicitly notes: “Your interactions with AI features can train AI models.” It seems this encompasses images shared with Meta AI through the Ray-Ban smart glasses; however, Meta still refused clarification.
According to spokespeople, the platform’s terms explicitly state that users consent to having their shared photos, including facial features, analyzed using Meta AI upon agreement.
Meta settles a high-profile court case related to its use of facial recognition technology in the workplace. The “Tag Ideas” Facebook feature, launched in 2011, underwent significant changes by 2021 when it became explicitly opt-in. As a result, billions of users’ biometric data were deleted, marking a notable shift in the platform’s handling of user information. Notably, several picture options from Meta AI have been experiencing issues with non-launching functionality in Texas.
The corporate reveals that, by default, it stores all transcriptions of your voice conversations with Ray-Ban Meta to train future AI models, leaving users in the dark about their data’s usage. Precise voice recordings offer an option to opt out. Upon initial login to the Ray-Ban Meta app, users have the option to permit or decline voice recording usage for training Meta’s AI models.
Tech giants like Meta and Snap are driving the development of smart glasses as a novel form of computing. The majority of these wearable camera devices are equipped with facial components worn by individuals, primarily dependent on artificial intelligence (AI) for operation. The debate surrounding privacy concerns has been reignited with the advent of Google Glass, echoing earlier discussions that date back to its initial announcement. According to a report by 404 Media, certain school college students are being asked to provide detailed information about anyone they encounter, including their title, topic of conversation, and phone number.
