What the FTC's Personalized-Pricing Orders Ask Delta, Uber and Walmart
Delta, JetBlue, Instacart, Walmart and Target have made public denials of personalized pricing. The FTC's orders skip the promises and ask each company to count sales per method, under oath.
A Delta spokesperson put it flatly to Business Insider that there is "no fare product Delta has ever used, is testing or plans to use that targets customers". That was Wednesday, Oct. 28, the day the Federal Trade Commission announced the orders it was sending to Delta and eight other companies under Section 6(b) of the FTC Act. The FTC approved them 2-0, and each recipient has 45 days from receipt to file a sworn report.
The order does not ask Delta that question. It asks Delta to count: for every method that uses personal information to set a consumer price, how many sales, and how many dollars, ran through it. Delta's sentence is a promise. The order wants a ledger.
That gap is the story: the denials say no individualized offers, while the order is a measurement exercise in which a sworn zero tells the FTC as much as a sworn number. Five of the nine recipients have made public statements denying some form of personalized pricing, though the wording differs by company. For United, Lyft, Uber and DoorDash, no statement had been reported as of Oct. 29.
| Company | Sector | What it said publicly (as reported) |
|---|---|---|
| Delta | Airline | No fare product it has ever used, is testing or plans to use targets customers with individualized offers based on personal data |
| JetBlue | Airline | Does not use personal information or browsing history to set individual prices; fares set by demand and seat availability |
| United | Airline | No statement reported |
| Instacart | Delivery | Does not use personal information to set prices, which retailers set; shopper pay based on job-related factors |
| Walmart | Retail | CEO said last month it does not use personal information such as income or shopping history to set prices and won't |
| Target | Retail | Oct. 15 fact sheet: does not use personal information to raise an item's price |
| Uber, Lyft, DoorDash | Ride-hail / delivery | No statement reported |
Statements as reported by AP, Reuters, Business Insider and PYMNTS through Oct. 29, 2026.
Delta speaks of individualized offers, JetBlue of individual prices and Target of raising an item's price. Instacart and Walmart say personal information is not used to set prices. As reported, none addresses how the company sorts people into categories, what its A/B tests since 2021 found, or which workers are shown which jobs, apart from Instacart's line on shopper pay. That is not evasion; the statements and the order cover different ground, and only a sworn answer closes the distance.
The order asks for counts, not promises
The order's 23 specifications are blunt about method. Specification 7 asks each company to identify every method that uses personal information as an input to a consumer price, the tools behind it (including AI), and whether each input varies by individual or only by category.
Specification 9 asks, by product and year, for total sales and revenue and for the number of sales and revenue that ran through each such method. A company that never used personal information would presumably report zero. One that did reports a share of its business.
Personal information covers location, device data, purchase and browsing history, income status and willingness to pay, plus anything inferred from them. A method is any formula, model or rule set, including sorting people into groups.
Workers are in the order too. Specifications 11 through 13 ask which workers are offered which jobs, in what order, with which jobs hidden or filtered. Specifications 14 through 16 cover worker pay, including the share of each consumer price the company keeps. For a driver, pay is not just a rate; it is also which jobs appear on the screen.
Other specifications ask how consumers and workers are grouped into categories (17), where each data type came from (18), and every A/B test since Jan. 1, 2021, with results (20). Specification 22 asks for the exact wording, location and dates of what the company tells consumers and workers, down to the URL.
"Companies know how to use language that is technically accurate but not informative."
Lauren E. Willis, law professor, Loyola Marymount University, to Consumer Reports
That is why Specification 22 matters. With the exact words in hand, the FTC can compare what companies tell customers with what they say under oath.
A study can compel answers but does not itself penalize pricing
None of this is an accusation. Hunton Andrews Kurth, a law firm, notes that receiving a 6(b) order does not establish wrongdoing, and the FTC's own release says the section lets it run broad studies with no specific law enforcement purpose. The statute lets the agency require reports under oath and then "make public from time to time such portions of the information obtained by it hereunder as are in the public interest," and recommend legislation to Congress. Reuters reports the agency will publish a study for regulators and lawmakers. A company that refuses can be taken to federal court, and the order warns of penalties for failing to file or filing a false report. None of that penalizes how a ticket was priced.
The order also says confidential commercial information will be reported in aggregate or anonymous form. The likely product is a pattern, not nine grade cards. No source gives a date for the report.
Chairman Andrew Ferguson framed the purpose in his own post on X on Oct. 28, asking why "similarly situated delivery drivers working identical routes" are paid "wildly varying amounts."
Post by @AFergusonFTC
He also posted that the FTC "has already issued an enforcement statement explaining that it is illegal to charge a personalized price without clearly informing the consumer that the price is personalized and the data on which the price is based." The Aug. 19 statement is a proposal, approved 2-0, and says undisclosed use of personal data to personalize prices could violate the FTC Act's ban on unfair or deceptive practices. Morgan Lewis, another law firm, notes a policy statement is neither a rule nor a binding legal determination.
The orders also replace an earlier effort. On Jan. 17, 2025, the agency voted 3-2 to release staff findings in January 2025 from a study of what the agency then called "surveillance pricing." That study began with 6(b) orders to eight companies that sell pricing tools, not to the businesses that set prices. The FTC now says that study looked only at intermediaries, not at the businesses that set prices. It took 649 days to get from those findings to these orders.
Those findings rested on vendors' documents and hypothetical examples, such as a consumer profiled as a new parent shown pricier baby thermometers first. Unlike that study, these orders go to the businesses that set prices.
Where the law already reaches a price
Until the report arrives, protection comes from state law, and it depends on where you shop. New York's Algorithmic Pricing Disclosure Act has applied since Nov. 10, 2025. A business that sets a price with personalized algorithmic pricing using a consumer's personal data must show this sentence: "THIS PRICE WAS SET BY AN ALGORITHM USING YOUR PERSONAL DATA." Only the state attorney general enforces it, at up to $1,000 per violation.
The exemptions matter to riders. Location data a for-hire vehicle uses to calculate a fare from mileage and travel time is exempt, so a fare computed that way would not carry the label. The law is also being fought. A federal judge dismissed the National Retail Federation's First Amendment challenge on Oct. 8, 2025; the Second Circuit heard the appeal on Monday, Oct. 19, 2026, and has not ruled.
Maryland's Protection From Predatory Pricing Act took effect Oct. 1, 2026. It bars food retailers (stores of at least 15,000 square feet selling tax-exempt food) and delivery apps from using personal data to set higher food prices for individual consumers. California has no such statute: AB 2564 failed this year, though its attorney general has treated the practice as a possible privacy issue. Connecticut's ban starts July 1, 2027, and New Jersey's, with a private right of action, on Aug. 1, 2027.
Our own report, Dollar General's Four Test Buys: 30 Price Errors, 21 Against the Shopper, is about price errors. Personalization is about two shoppers being shown different posted prices.
What this looks like from a seat, a car or a cart
Delta's president said on a July 10, 2025 earnings call that about 3% of domestic ticket prices were set using AI, per Newsweek, and Democratic senators wrote that "Prices could be dictated not by supply and demand, but by individual need." Delta says its AI-assisted dynamic pricing reflects market forces and does not target individuals. AI setting a fare is not a fare set for you, and that line is what is disputed.
At the cart, the evidence is real but narrow. A Dec. 9, 2025 investigation by Consumer Reports and two other groups analyzed 193 shopper submissions in four cities and found nearly three-quarters of tested items showed different prices to different shoppers. A dozen Lucerne eggs at a Washington, D.C. Safeway ranged from $3.99 to $4.79. Instacart acknowledged price tests, called them randomized and "negligible," and announced it was ending them Dec. 22, 2025. Reuters and Instacart both say these were random experiments, not targeting.
Behind the wheel, the evidence is thinner. Business Insider reports Uber shows drivers payouts based on factors such as demand, not a set per-mile rate. An Oxford and Worker Info Exchange study found hourly pay for 258 UK Uber drivers fell from £22.20 to £19.06 (inflation-adjusted, on a tribunal working-time measure) after Uber introduced dynamic pricing in 2023. Lead author Reuben Binns said: "The higher the value of the trip, the more of a cut Uber takes." But the 258 are self-selected members of roughly 100,000 drivers, and the authors say the design cannot isolate cause. It concerns dynamic pay, not proven use of any driver's data.
No source tests whether any step below changes a price. Consumer Reports' guidance is to look for disclosures like New York's label, remember that loyalty apps personalize discounts from purchase history, and limit location access and account details, since billing ZIP code and geolocation are data sources. In this writer's suggestion, comparing a price logged out and logged in is a way to notice differences, not a proven fix, and it will not catch random tests like Instacart's. UK drivers used data access requests for the Oxford study; no source establishes a U.S. equivalent.
Delta, JetBlue, Instacart, Walmart and Target have each made public denials, in differing words, of personalizing prices. Under oath, in roughly mid-December, each will have to show it in counts. What no source says is when anyone outside the FTC gets to see the numbers.
