India's AI-driven food bank access program represents a case study in what happens when efficiency arguments override population reality. The system uses facial recognition to verify that the person collecting food is the registered beneficiary — a design that assumes bodies are stable, legible, and recognizable across time. Pregnancy violates that assumption on every axis: weight distribution shifts, facial geometry changes, and the recognition confidence threshold drops below what triggers access. No one in the system's design chain made a deliberate choice to exclude pregnant women. The exclusion followed from…