Meta’s Silence on Training AI with Smart Glasses Photos Raises Privacy Concerns

Meta’s AI-powered Ray-Bans, featuring a discreet front camera, capture photos not just on command but also through AI triggers like certain keywords. This leads to a substantial collection of images, raising concerns about their privacy and subsequent use by Meta.

Meta’s AI-powered Ray-Bans have introduced a new level of convenience with their discreet front camera, capable of taking photos not just when prompted but also when certain AI features are activated. This means a significant number of images are collected, both intentionally and passively.

When asked if these images are used to train AI models, Meta remained evasive. According to Anuj Kumar, a senior director at Meta, “We’re not publicly discussing that.” Similarly, Meta’s spokesperson, Mimi Huggins, stated, “We’re not saying either way.”

The ambiguity surrounding Meta’s data usage policies is particularly alarming given the glasses’ new feature, which streams images in real-time to an AI model, enabling it to provide low-latency, natural responses to user queries. This could involve numerous images being uploaded to the cloud without the user’s full awareness.

For instance, asking the smart glasses to scan a closet for outfit suggestions results in numerous photos being taken and stored. The ultimate fate of these photos remains unclear, as Meta has not committed to keeping them private.

The presence of a camera on the Ray-Ban Meta glasses is reminiscent of the backlash faced by Google Glass, where people were uncomfortable with the constant recording potential. It’s surprising that Meta has not assured users of their privacy by stating that all photos and videos will be confined to the user’s device.

Meta has already declared that it trains its AI models on publicly available data from Instagram and Facebook. This broad definition of publicly available data has been a point of contention. However, the images captured by smart glasses do not fit this category and should be treated with more confidentiality.

Other AI model providers, like Anthropic and OpenAI, have clear policies that they don’t train on user data. Meta’s reluctance to clarify its stance on data privacy puts it at odds with these companies, raising further concerns about user data security.

The lack of transparency from Meta regarding the use of photos collected through their AI-powered Ray-Bans is troubling. Users deserve clear answers about how their data is being used, especially when it involves personal images captured unknowingly. The silence from Meta only amplifies these concerns.

About The Author

Emmanuel Kesse

See author's posts