The disclosure that Meta's smart glasses were transmitting recorded data to workers in Nairobi for manual labeling highlights a critical flaw in AI privacy claims. While marketed as 'privacy-first,' the underlying AI training pipeline relied on the mass exploitation of human labor and the subversion of user consent. The subsequent firing of 1,108 whistleblowers underscores the ethical and security risks of centralized AI data processing. This event serves as a warning: 'AI-on-the-edge' often conceals a 'Human-in-the-loop' reality that is vulnerable to both data leaks and ethical collapse.