AI Glasses Voice Privacy Notice

Effective April 29, 2025

The voice services available to you depend on the Meta Wearable Product and your location. The Meta AI service on Ray-Ban | Meta AI glasses allows you to do things like get information, capture photos and videos, play music, and call and message using your voice.

Voice interactions are things you say while using the voice services on Meta Wearable Products, such as voice controls on Ray-Ban Stories or the Meta AI service on Ray-Ban | Meta AI glasses. Voice interactions include: things you say when you use a wake-word (for example, if your AI Glasses are playing music, and you say “Hey Meta, pause the music”); when you don’t use a wake-word for certain interactions (saying “pause music” if your AI Glasses are playing music, or “play message” after we tell you that you have a new message); or if you inadvertently activate our voice services, as well as any background sound that happens when you are interacting with our voice services.

The Meta AI service on your Ray-Ban | Meta AI glasses will activate when they detect the wake word (for instance, “Hey Meta” or another wake word, if available) so that it can respond to your voice. You can also activate the Meta AI service on your glasses to respond to your voice by pressing and holding the touchpad, if the touchpad has been designated to activate the Meta AI service on your glasses in your app settings. During a response you can also issue additional commands like “stop” or “cancel.”

When you use voice services, like voice controls on Ray-Ban Stories or the Meta AI service on Ray-Ban | Meta AI glasses, we process audio recordings of your voice interactions to create text transcripts of what you say, as well as data related to your voice interactions, to respond to your requests. Related data includes information such as the hardware version of your glasses and the length of the audio related to your interaction.

We primarily use recordings and transcripts to process and carry out your requests. Text transcripts and audio recordings of your voice interactions are stored by default to help improve Meta’s products. This includes improving the Meta AI service (and related technologies) and features, and training and improving our speech technologies. For example, audio recordings are stored and used to help Meta’s products better understand and fulfill requests. We use machine learning and trained reviewers to process this data to improve Meta’s products. You can access and delete recordings and related transcripts in the Meta AI App. Stored audio recordings and related data are used to improve Meta’s speech technologies (for example, our speech-to-text technologies) with the help of trained reviewers, as described below.

Trained reviewers are useful in evaluating how well Meta’s products respond to your voice. For example, reviewers can evaluate if the glasses actually played the next song following a command like “Hey Meta, next song,” and help improve Meta’s voice services to respond more accurately over time. Making sure our products can understand and respond to a wide range of speech patterns, phrases, local dialects, and accents is important to us. Trained reviewers are particularly helpful in identifying voice interactions that machines are likely to misinterpret.

We will store voice recordings even if you unintentionally activate a voice interaction. If our systems detect that you didn’t intend to activate a voice interaction, we will label these voice interactions as “false wakes” or misactivations, and delete them within 90 days of detection. Voice transcripts and stored audio recordings are otherwise stored for up to one year to help improve Meta’s products. You can delete stored voice recordings and transcripts sooner in your App. After deletion, they will no longer be used to improve Meta products. Deleting a recording and transcript will also delete the data associated with that interaction.

We primarily use recordings and transcripts to process and carry out your requests. We also use text transcripts and related data to help improve Meta’s products. You can choose to store your voice recordings to help improve Meta’s voice controls and Meta products on Ray-Ban Stories. When storage is enabled, audio recordings of your voice interactions are stored to help Meta’s products better understand and fulfill requests. We use machine learning and trained reviewers to process this data to improve, troubleshoot and train our voice controls and other Meta products. You can turn voice storage off in your settings at any time. Even with storage off, you can still use voice controls. You can also access and delete stored information in your settings.

Storing your voice recordings will help Meta’s products better understand and fulfill requests, as well as respond to a wide range of speech patterns, phrases, local dialects, and accents. For example, people that speak a regional dialect and have storage of their voice recordings enabled will help Meta’s products better understand and more accurately respond to requests from people that speak that regional dialect.

Even if voice recording storage is turned off on your Ray-Ban Stories, you can still use voice controls. When storage of voice recordings is turned off, we will delete all voice recordings immediately after processing, which means you won’t be able to hear, view or delete this information. We also will not use your voice recordings to improve Meta’s products and voice recordings will not be reviewed by trained reviewers. Any voice recordings that occurred before you turned off storage will still be stored for up to one year, unless you delete them sooner. You can manage storage of your voice recordings in Settings.

We have a team of vetted and trained personnel who assist in reviewing stored audio recordings of your voice interactions, for purposes of improving Meta’s products. These individuals must comply with strict privacy and security requirements in their handling of this information. We use controlled and monitored systems to ensure that they access only the information required to complete product improvement training tasks. And we protect your privacy by changing the pitch of voice recordings when they are reviewed.

You can download stored transcripts and related data, like the serial number of your glasses or the general category of your voice interaction with the Download Your Information tool. You can access this tool from the App or on the web.