Kroger's Surveillance Pricing Harms Consumers and Raises Prices (epic.org)

Kroger has come under scrutiny for practices described as "surveillance pricing"—using vast troves of behavioral data (purchase history, app and browsing data, loyalty programs, and third‑party broker data) to infer sensitive attributes and tailor prices or offers to individual shoppers. Lawmakers raised alarms about electronic shelf labels (ESLs) and Cooler Screens’ digital displays with embedded cameras and sensors that can detect presence, dwell time, and potentially infer age/gender. Kroger denies using facial recognition to identify customers, but ESLs and in‑store sensors materially expand the reach of personalized pricing from online into brick‑and‑mortar environments. Examples cited (Target, Orbitz) show how platforms already adjust prices based on user signals, and ESLs let retailers change shelf prices in seconds, enabling rapid, location- and time‑dependent price discrimination. For the AI/ML community this is a warning: models that infer demographics or willingness‑to‑pay enable opaque, individualized price optimization that raises technical, ethical, and regulatory issues. Key implications include algorithmic price discrimination, privacy risks from sensitive attribute inference, cross‑channel profiling (app, beacon, camera), and lack of transparency about model inputs and outputs. The piece urges two policy fixes relevant to practitioners: enforce data minimization to limit inputs for predictive models and regulate facial‑recognition/deployment of sensorized displays. Engineers and researchers should prioritize privacy‑preserving modeling, explainability, and impact assessments when developing systems that monetize behavioral inferences.