Meta’s AI-powered Ray-Bans have a discreet camera on the front, to take photos not only when you ask them to, but also when their AI features trigger them with certain keywords like “look.” That means the smart glasses collect a ton of photos, both taken deliberately and otherwise. But the company will not commit to keeping these images private.
We asked Meta if it plans to train AI models with images of Ray-Ban Meta wearers, as it does with images from public social media accounts. The company would not say so.
“We’re not discussing that publicly,” Anuj Kumar, a senior director working on AI wearables at Meta, said in a video interview with britcommerce on Monday.
“That’s not something we normally share externally,” said Meta spokeswoman Mimi Huggins, who was also on the video call. When britcommerce asked for clarification on whether Meta is training with these images, Huggins responded: “We’re not saying any way.”
Part of the reason this is especially concerning is because of Ray-Ban Meta’s new AI feature, which will take many of these passive photos. Last week, britcommerce reported that Meta plans to launch a new streaming video feature for Ray-Ban Meta. When activated with certain keywords, the smart glasses will transmit a series of images (essentially live video) to a multimodal AI model, allowing it to answer questions about its environment in a natural, low-latency way.
That’s a lot of pictures, and they’re pictures that a Ray-Ban Meta wearer may not be aware they’re taking. Let’s say you ask smart glasses to scan the contents of your closet to help you choose an outfit. The glasses effectively take dozens of photos of your room and everything in it, and upload them all to an AI model in the cloud.
What happens to those photos after that? Meta won’t tell.
Wearing Ray-Ban Meta glasses also means you’re wearing a camera on your face. As we discovered with Google Glass, that’s not something other people are universally comfortable with, to put it lightly. So, you’d think it’s a no-brainer for the company that’s doing it to say, “Hey! All your photos and videos from your face cameras will be completely private and isolated on your face camera.”
But that’s not what Meta is doing here.
Meta has already stated that it is training its AI models on The public Instagram and Facebook posts of every American.. The company has decided that all of that is “publicly available data” and we may have to accept that. This and other tech companies have adopted a very broad definition of what is publicly available for training AI and what is not.
However, surely the world you see through your smart glasses is not “publicly available.” While we can’t say for sure that Meta is training AI models on the images from its Ray-Ban Meta camera, the company just wouldn’t say for sure that it isn’t.
Other AI model vendors have clearer rules about training on user data. Anthropo says it never trains the inputs or outputs of a client in one of your AI models. OpenAI says it too never trains user inputs or outputs through its API.
We’ve reached out to Meta for further clarification here and will update the story if they hear back.