How to stop Meta training its AI models on your smart glasses' visual data

There's a quiet irony in the fact that the most consequential privacy decision you'll make with a pair of $300 smart glasses is buried in a settings menu. The Engadget walkthrough (https://www.engadget.com/2269454/how-to-stop-meta-training-its-ai-models-on-your-smart-glasses-visual-data/) reveals that the real question isn't whether Meta's algorithms "see" your photos—it's whether human workers get to. That distinction matters because it exposes the labor layer that most AI discourse conveniently forgets: training data doesn't label itself. The "how" here is refreshingly concrete. A setting determines whether reviewers can access visual data captured by your glasses, which means the entire privacy calculus hinges on a binary switch most users will never locate. This is the fundamental design pattern of the surveillance economy: offer a control, bury it, and let inertia do the rest. The existence of an opt-out is not the same as meaningful consent, and the fact that this requires a how-to article at all is a damning indictment of Meta's default choices. What makes this story more than a consumer PSA is what it reveals about the AI supply chain. Every "smart" feature Meta ships—whether it's answering questions about what you