# Meta’s Ray-Ban Glasses: Default AI Settings Raise Privacy Concerns

## Meta’s Ray-Ban Glasses: Default AI Settings Raise Privacy Concerns

Meta has quietly updated the privacy policy for its Ray-Ban Meta smart glasses, sparking concerns over user data and AI training. The update, communicated to owners via email on Tuesday, involves enabling AI features by default, granting Meta broader access to user data for improving its AI models.

According to the updated policy, Meta’s AI will automatically analyze photos and videos captured with the glasses when specific AI features are active. Furthermore, the company will store voice recordings of users to enhance its products, leaving users with no explicit option to opt out of this data collection.

It’s important to note that the Ray-Ban Meta glasses are not continuously recording and storing all surrounding audio and video. Only speech following the “Hey Meta” wake word is processed and potentially stored. However, this data, including voice transcripts and recordings, can be retained for up to a year, as stated in Meta’s privacy notice on voice services for wearables.

Users who are uncomfortable with Meta utilizing their voice data for AI training will need to manually delete individual recordings through the Ray-Ban Meta companion app – a potentially tedious and easily overlooked task.

This shift echoes Amazon’s recent policy change affecting Echo devices, where all voice commands are now processed through the cloud, eliminating the option for local, privacy-friendly data processing.

The motivation behind these changes is clear: large tech companies like Meta and Amazon are eager to amass vast amounts of voice recordings. This data is invaluable for training generative AI models, allowing them to better understand and process diverse accents, dialects, and speech patterns.

However, this pursuit of improved AI comes at the expense of user privacy. The average user may not fully grasp the implications of using their Ray-Ban Meta glasses, potentially unknowingly contributing images and audio of themselves and others to Meta’s AI training datasets. The sheer volume of data required for these AI models makes user-generated content an appealing and readily available resource for these companies.

This move is consistent with Meta’s existing practices, where its Llama AI models are already trained on publicly shared posts from Facebook and Instagram. While the allure of advanced AI features is undeniable, Ray-Ban Meta users should carefully review their privacy settings and consider the potential implications of sharing their data with Meta. The onus is on users to proactively manage their privacy and ensure their comfort level with the level of data being collected.