Skip to content

How we test new products

In the next few years we’re going to introduce products designed to help people feel more present and connected to others and the world around them. In order for these devices to be helpful for you in the real world, we need to test them in the same types of environments you’ll use them in, doing the same everyday activities you might do. To do that, we’re taking these new prototypes outdoors and into public spaces, beginning in the US. Some will take on new forms–like smart glasses–and include cameras and microphones.

In the spirit of continuing to innovate responsibly and openly, we want to share more about our approach to product development and testing, and the types of information we’ll collect and process during testing.

Our Approach

  • In following our Responsible Innovation Principles, testing helps ensure our products are designed in an inclusive, privacy-conscious way.
  • All testers must agree to and adhere to specific operating guidelines that demonstrate care for bystanders and our community. For example, testers won’t wear devices in areas with an expectation of privacy, such as bathrooms, places of worship, or doctor’s offices.
  • All product testing is conducted by Facebook employees, authorized contractors, and development partners.

Prototype Capabilities

  • Prototype design and style will vary, and will include novel devices, like smart glasses.
  • In the case of our smart glasses, prototypes allow testers to capture photos and video (including audio) hands-free. While the cameras are in use, an LED on the device lights up to let people nearby know when a recording or capture is taking place.
  • Photos and video are only captured when a tester takes an explicit action. For example, when they use a voice command or manually trigger the device. This means it will be clear when something is happening–nothing is being captured in secret.

Data and Privacy

  • When testers do take photos or record video (including audio), it’s only to test and improve the device’s features and capabilities. As with photos and videos taken on your phone, bystanders could be in the background.
  • Testers may send media to Facebook servers in order to help with product feedback or bug reporting, but no photos or videos are used to train Facebook’s facial recognition models. Otherwise, the media stays on the testers’ device.
  • Testers are instructed not to publish media to their external social channels.
  • We instruct testers to avoid using voice commands in situations where the device may accidentally capture conversations of people nearby, or when bystanders may be having a private conversation.
  • If a tester chooses to store voice commands, we will analyze and review those interactions to improve, troubleshoot, and train our speech recognition features.

We will update this page as our work progresses, as new prototypes might be subject to different terms and policies. Learn more about our commitments to responsible innovation here, and to learn more about Facebook’s Data Policy, visit here.