Meta has changed the default settings on its Ray-Ban smart glasses, meaning voice recordings and photos may now be stored and used to train its AI, unless users take action to stop it.

The update, which came into effect in late April, enables AI features by default. When triggered by a wake word like “Hey Meta”, the glasses can now record and store voice interactions for up to a year. These recordings may be reviewed by humans and used to improve Meta’s AI models, with no option to opt out, and only a manual delete option in the app.

Meta has also confirmed that photos and videos taken with AI features active may be analysed too, potentially putting bystanders’ faces into AI training data without their knowledge. The company says this helps its systems understand different accents and commands more accurately.

The policy seems to follow Amazon’s recent shift with Echo devices, where cloud processing now replaces local handling, thereby raising fresh concerns about always-on surveillance in consumer tech.

Businesses using AI wearables should review settings immediately, disable voice storage if possible, and ensure staff are trained to avoid capturing sensitive information.