Meta Platforms Inc. has updated the privacy policy for its Ray-Ban Meta smart glasses, expanding the company’s ability to collect, store, and use consumer data for artificial intelligence development. The update, announced to device owners via email on Tuesday, means AI-powered features will now be enabled by default, including voice recording and image analysis capabilities.
According to a report from The Verge, the tech giant will begin using audio, photo, and video data captured through the smart glasses to help train and improve its AI models. Notably, users are not given a global option to opt out of the data collection — a shift that has raised privacy concerns among advocates and experts.
Smart Glasses, Smarter AI — and More Data Collection
Ray-Ban Meta glasses, released in partnership with the iconic eyewear brand, offer a range of features such as hands-free photo and video capture, live streaming, voice-activated assistance, and now, integrated AI capabilities. The latest policy update allows Meta to analyze media captured by users when AI features are active, and to store voice recordings triggered by the wake command “Hey Meta.”
The company says the glasses are not constantly recording and do not store ambient sounds. Instead, they store only speech initiated by the user after the wake word is spoken. However, these voice interactions, once recorded, may be retained for up to one year, according to Meta’s updated privacy policy on voice services for wearable devices.
Customers who prefer not to have their voice data used to train Meta’s AI must manually delete individual recordings using the Ray-Ban Meta companion app. There is no built-in setting that allows users to opt out entirely from contributing to the company’s AI training processes.
Comparison to Amazon Echo Changes
The policy shift mirrors a recent move by Amazon, which announced in March that all Echo voice commands would now be processed via the cloud. That decision removed the previously available option to process commands locally on the device, which had been considered more privacy-friendly. Like Amazon, Meta is eager to harvest large quantities of voice and interaction data to enhance its generative AI systems.
These massive data sets — especially those containing natural, unscripted human speech — are crucial for improving AI models’ ability to understand a range of languages, dialects, accents, and conversational patterns. The tradeoff, however, is greater exposure of private user data to corporate servers and algorithms.
ADVERTISEMENT
A Broader Trend Toward Data-Driven AI
As Meta continues to develop its Llama family of generative AI models, the company has already drawn scrutiny for the ways in which it collects training data. Public posts from U.S.-based Facebook and Instagram users are currently being used to train these models, a practice that was quietly introduced and updated in previous versions of Meta’s privacy terms.
The use of data generated by wearable devices represents an even more personal and potentially invasive form of collection. For example, a user taking a photo of a loved one or recording a video in a private setting could inadvertently contribute that content to Meta’s AI training datasets. These images, voices, and interactions could be analyzed by machine learning systems for pattern recognition and response modeling — often without the express awareness or consent of those being recorded.
Privacy Advocates Raise Concerns
Digital rights groups and privacy advocates have long warned about the implications of ambient surveillance technologies and opaque data usage policies. Critics argue that most users are not fully informed about how their data is being used, and that the burden of manually deleting individual recordings places an unreasonable responsibility on consumers.
“The shift toward default data collection and AI training reflects a dangerous erosion of user control,” said Clara Montoya, a senior analyst with the nonprofit Privacy First. “These smart devices are becoming Trojan horses for surveillance and data mining under the guise of convenience.”
Meta Defends Its AI Vision
Meta has defended its AI strategy by highlighting the importance of real-world data in building effective, responsive, and inclusive AI tools. In its public documentation, the company emphasizes that data collection is conducted with safeguards in place, and that user information is anonymized where possible.
Still, the lack of a comprehensive opt-out mechanism and the increasing use of personal media for commercial AI development continue to fuel criticism.
ADVERTISEMENT
What’s Next for Wearable AI
The growing integration of AI into consumer hardware reflects a broader industry trend. Companies like Meta, Amazon, Apple, and Google are all racing to dominate the market for wearable assistants and multimodal AI interfaces — tools that combine voice, vision, and context to interact more naturally with users.
Meta is expected to unveil new AI features at its upcoming developer conference in May, with a particular focus on expanding the functionality of its Ray-Ban Meta glasses and Portal smart home devices.
For now, users are encouraged to review their privacy settings within the Meta ecosystem and understand how their data may be used. As wearables become smarter, the lines between convenience, innovation, and surveillance continue to blur.
