Privacy and Pricing: Separating Paranoia from Truth
When I flip through my TikTok feed, I find the occasional story where a person accuses a retailer of reading your license plate, using facial recognition, detecting your phone’s WiFi and Bluetooth and deciding whether you can afford to pay more. Add digital shelf labels, and the story becomes a price increase as you walk down the aisle or that the price at checkout is higher than the price marked on the shelf.
The technologies make that scenario plausible. They do not establish that it happens. The useful question is where the evidence stops and speculation begins. The purpose of this article is not to prove or disprove that the practice occurs rather it is to show examples and potential legislative changes to protect users from abusive practices. I usually do not provide citations or links in my articles but since this article addresses accusations, it seems appropriate.
What the Walmart allegation gets right — and leaves unproven
Walmart’s license plate reader notice confirms collection for security, theft and fraud prevention, parking enforcement, and safety. Its customer privacy notice describes device identifiers, permission-dependent location collection, in-store Wi-Fi/Bluetooth capabilities and personalized promotions. Those are substantial privacy disclosures. They do not establish that these inputs determine individualized prices.
Facial scanning also needs context. Virtual eyeglass try-on, capturing checkout images and identifying a shopper are different activities. A 2022 lawsuit alleged Walmart used Clearview facial recognition, but a complaint is an allegation, not an adjudicated finding or proof of pricing abuse.
Walmart says its digital labels contain no cameras or microphones and collect no customer data. On September 25, 2026, CEO John Furner explicitly denied prices based on identity, income, shopping history or urgency.
These statements deserve inclusion, but they are not independent audits. The defensible conclusion is that the alleged surveillance-to-price connection remains unverified.
Comparison with other companies
Company or system
Documented practice
Strength and limits of evidence
Shopping history and other information inform personalized offers.
Company disclosure; supports offer personalization, not biometric shelf surcharges.
Rewards uses personal and transaction data for offers.
Company disclosure; does not establish individualized higher menu prices.
Acknowledged different same-item prices; announced tests ended December 2025.
Deployed experiments confirmed; personal-data assignment denied.
Safeway and Target on Instacart
Included in observed 2025 price differences.
Get Jason Hede’s stories in your inbox
Join Medium for free to get updates from this writer.
Original field tests; do not establish the retailers’ own in-store practices.
Confirmed an AI pricing pilot using aggregate data.
Company account; individual personal-data fares denied.
Announced demographic video analytics for offers and ads in 2019.
Historical proposal; later reporting attributes pilot termination and a facial-recognition denial to Kroger.
Orbitz, Cheaptickets, Travelocity, Expedia and Home Depot
Academic tests found several forms of personalization.
Controlled 2014 research; mechanisms differed and findings do not establish current deployments.
Different mechanisms can produce different prices. Dynamic pricing responds to conditions such as demand or inventory. Surveillance pricing uses customer-specific information. Coupons change the effective price; steering changes which products appear. Randomized tests can produce differences without targeting someone’s income or identity.
An online-versus-store mismatch alone cannot tell us which mechanism caused it.
What studies and tests actually show
The FTC’s January 2025 staff study described intermediaries offering data-supported pricing and promotion tools. Its public examples were hypothetical, drawn from anonymized material. The reported minimum of 250 clients was not a count of retailers proven to impose personal surcharges. Commissioners Ferguson and Holyoak criticized the preliminary release, arguing for more complete findings.
The Instacart investigation supplies a concrete test. Researchers observed same-item differences reaching 23%. However, measured demographic and shopping-history associations were small and not statistically meaningful. The sample was not nationally representative.
Instacart said the assignments were randomized, disputed personal-data pricing and later ended the tests. The investigation demonstrated price variation; it did not demonstrate that personal characteristics caused it.
Privacy advocates also question uneven discount eligibility. EPIC’s analysis argues that this deserves scrutiny even without facial recognition. A coupon can provide real savings while making access to those savings depend on profiling.
What legislation changes
U.S. protection varies by jurisdiction and practice. The FTC’s August 2026 proposed policy warns that undisclosed personalized pricing could violate existing consumer-protection law. It also acknowledges that the agency cannot ban the practice in every circumstance.
New York requires disclosure near covered personalized prices. Maryland’s law, effective October 1, 2026, restricts specified food-pricing practices and includes exceptions for promotions, loyalty programs and other circumstances. Colorado’s proposed ban was vetoed.
These approaches illustrate the gap between requiring transparency and restricting use. A disclosure may expose profiling without preventing it.
For security professionals, the question is whether loss-prevention, identity and commercial systems share data across approved purpose boundaries. NIST recognizes that processing can cause privacy harm, including economic loss, even without a breach.
We should ask which inputs influenced an offer, who authorized their use and whether the decision can be reconstructed. Skepticism should apply to alarming accusations and reassuring corporate statements alike.
