WASHINGTON — Increasing the likelihood members and consumers will pay different prices for the same cars, the growing use of artificial intelligence to set individualized prices based on consumers’ personal information is drawing scrutiny from federal regulators and lawmakers, including concerns about what the practice could mean when Americans shop for cars — and for credit unions that finance those purchases.
The practice, often called “surveillance pricing” or “personalized pricing,” uses algorithms and consumer data to estimate what an individual is willing to pay and potentially offer different consumers different prices for the same product or service.
The Federal Trade Commission has found that companies offering algorithmic pricing technology can use detailed information including consumers’ location, demographics, browsing behavior and shopping history to tailor prices and promotions.
The FTC has specifically identified automobiles as an area where surveillance pricing could pose heightened concerns because consumers may invest considerable time in the purchasing process before learning the vehicle’s actual cost.
There is no clear evidence, however, that major used-car retailers are currently taking an individual shopper’s income and using AI to change the advertised price of the same vehicle. Instead, regulators are raising concerns about technology that has the capability to use personal information — including data that could reveal or help infer a consumer’s financial circumstances — to determine willingness to pay.
AI Can Build a Consumer Profile
When the FTC launched an investigation into surveillance pricing in 2024, the agency said companies were offering technology that combined advanced algorithms and AI with information including location, demographics, credit history and browsing or shopping history to categorize consumers and establish targeted prices.
The FTC subsequently found pricing intermediaries could access direct and inferred consumer data and use that information to influence the prices consumers see. Even mouse movements, abandoned online shopping carts and previous searches can become inputs into pricing systems, according to the agency.
That means an algorithm would not necessarily need to know that a car buyer earns, for example, $150,000 annually.
Instead, data including ZIP code, credit characteristics, purchasing patterns, browsing behavior and other demographic and behavioral information could potentially allow an algorithm to infer the consumer’s economic circumstances or willingness to pay.
The FTC said advances in machine learning have made collecting and analyzing large amounts of such information significantly easier and warned consumers could potentially encounter surveillance pricing when purchasing anything from groceries to a house or car.
FTC Considering New Pricing Policy
The issue has taken on new urgency.
The FTC in August proposed an enforcement policy addressing personalized pricing, which it defines as using personal data to establish prices according to what a company believes an individual consumer is willing to spend.
FTC Chairman Andrew Ferguson said consumers seeing a listed price generally expect other consumers to see the same price rather than a retailer’s estimate of how much they personally are willing to pay.
The commission said it does not have authority to prohibit personalized pricing in all circumstances but warned that businesses failing to disclose how personal information is being used to establish prices could violate the FTC Act or other laws.
The FTC is accepting public comments on the proposal through Sept. 18.
Congress also is investigating the issue.
House Oversight Committee Chairman James Comer, R-Ky., said Wednesday the committee is continuing an investigation begun in March into the use of AI for surveillance pricing.
Comer said technology exists to track consumer behavioral information and build profiles using geolocation, demographics, browsing and purchase histories, device type, battery life and even mouse clicks.
The concern, according to Comer, is whether algorithms can use that information to determine an individual consumer’s maximum “willingness to pay.”
Auto Market Has History of Pricing Disparities
Academic research suggests the issue could be particularly relevant to automobile sales.
A National Bureau of Economic Research study examining automobile transactions found that traditional in-person transactions provide salespeople with information that can help them assess a customer’s willingness to pay.
Researchers found relationships among vehicle prices and demographic factors including income, education and consumers’ costs of searching for alternatives. They also found that moving transactions online removed some of the information salespeople could use to assess individual buyers and reduced certain pricing disparities.
An earlier study of used-car markets in Wisconsin, Iowa and Minnesota found poor consumers paid more for used vehicles than non-poor consumers and received less redress when defects were discovered.
Those differences persisted even after researchers controlled for factors including vehicle model and age. The research predates modern AI and does not establish that algorithms are responsible for income-related disparities in today’s auto market.
Implications for Credit Unions
For credit unions, personalized vehicle pricing could introduce a new variable into an already complex indirect and direct auto lending market.
If two members interested in an identical vehicle were quoted different prices because an algorithm determined one had a greater ability or willingness to pay, those differences could flow directly into financing.
A higher purchase price could mean a larger loan balance, higher loan-to-value ratio, larger monthly payment or greater required down payment. It also could affect underwriting and potentially expose members to higher borrowing costs over the life of the loan.
That could leave credit unions financing a price that was determined not simply by the vehicle’s market value, condition, mileage and dealer margin, but partly by an algorithm’s assessment of the individual member.
The issue could become particularly significant for credit unions that emphasize financial well-being and fair access to credit if algorithms use variables correlated with income or other demographic characteristics to establish vehicle prices.
Additional Warnings
The FTC has warned more broadly that surveillance pricing could create opportunities for deceptive advertising, illegal collusion and civil rights violations.
For now, the distinction is important: regulators have documented technology capable of using extensive personal information to customize consumer prices, and the FTC has specifically cited automobile purchases as an area of potential concern. But evidence has not emerged establishing that major used-car dealers are systematically using individual shoppers’ incomes to set AI-generated prices.
As those systems become more sophisticated, however, credit unions may have reason to pay attention not only to whether a member can afford the vehicle being financed, but also to how the price of that vehicle was established in the first place.
