The Truth-in-Lending Problem Hiding in Your Dealership’s Website

A shopper browsing your inventory at midnight asks the website assistant what the used SUV would cost a month and whether it still has warranty coverage. The bot, built to keep the conversation moving, answers: "around $399 a month" and "yes, it’s still covered." Neither number went past your finance desk. Neither claim matched the Buyers Guide on the windshield. By morning the shopper has a screenshot, an expectation, and the beginning of a complaint, and the dealership owns every word the website said.

Auto retail is one of the most heavily regulated sales environments in the country, and a general-purpose chatbot does not know that. It was hired to capture leads off the inventory pages. What it actually does is advertise credit terms and make warranty representations the dealership is accountable for, and the regulators who police those statements do not care that a bot said them.

A Payment Quote Is an Advertised Credit Term

When a shopper asks "what would my payment be," the honest answer depends on the price, the down payment, the term, the rate, and the buyer’s credit, none of which the website knows. A generic bot answers anyway, because answering is its whole purpose. The "$399 a month" it produces is not small talk. The moment a specific payment or rate goes out, you are advertising credit, and Regulation Z’s advertising rules require specific disclosures to travel with that number. A bare monthly figure with none of the required terms is the kind of triggering-term violation that has drawn FTC and state attention to dealers for years, except now it is sitting in a chat transcript with your name on it.

"Still Under Warranty" Is a Representation, Not a Reassurance

The warranty answer is its own trap. Federal law draws a bright line between a used car sold "as is" and one sold with a warranty, and the FTC’s Used Car Rule requires the Buyers Guide to state which one applies. When a chatbot tells a shopper the vehicle is "still covered," it can directly contradict the "as is" disclosure on the glass, and a written promise about coverage can pull the sale under the Magnuson-Moss Warranty Act. The dealership did not decide to extend a warranty. A website widget did, in a sentence, and the buyer is entitled to rely on it.

Vehicle history is the same hazard at a different address. A bot that volunteers mileage or condition details it cannot verify wanders toward the federal odometer disclosure rules and toward a misrepresentation claim if the detail is wrong. None of this is what the dealer set out to automate.

The Advertised Price Carries Its Own Exposure

Price is where it gets worse, because a chatbot rarely stops at the sticker. Asked for an out-the-door number, it estimates, folding in or leaving out tax, title, and doc fees it has no way to calculate for a specific deal. A figure that turns out lower than the real total is the kind of gap that draws deceptive-advertising scrutiny under the FTC’s authority over unfair and deceptive practices, and a shopper who drove in on the strength of the website’s number is primed to feel baited when the desk quotes something higher. Trade-in answers carry the same risk in reverse: a bot that floats what a trade "should be worth" sets an expectation no appraiser validated, and the negotiation starts from a number the dealership never authorized. None of this is hypothetical edge-casing. It is the ordinary path of a price conversation, run by a system that treats every question as one more thing to answer helpfully.

"Will Not" Is a Suggestion. "Cannot" Is an Architecture.

Here is what trips dealers up. A shopper who wants a number does not ask once and accept "please contact our sales team." They push. "Okay, ballpark, what would something like this run a month?" A chatbot steered only by a prompt eventually answers the version of the question it was not specifically warned about, because being helpful is its default and a prompt is only a polite request to suppress that default.

That is the gap between a tool that is told not to quote financing and one that is built so it cannot. Bolting a disclaimer onto a system whose entire job is to produce confident answers does not change what the system does when a motivated shopper leans on it. It just hands the plaintiff’s attorney something to read into the record.

Who Owns the Answer

Strip away the software and the exposure is familiar. A salesperson who quotes a fake payment or promises phantom coverage creates liability, and the dealership trains and supervises against exactly that. When an unsupervised automated system does the same thing on the website, the liability does not disappear. It lands on the store, now with a written record, no F&I review, and regulators who have spent a decade making clear that the dealer is responsible for what it advertises.

The dealers who get burned are not the ones who modernized their websites. They are the ones who dropped in a generic chatbot, assumed "helpful" and "compliant" were the same thing, and discovered the difference in a demand letter. The answer is not to strip AI off the site. It is to run one that knows a payment quote and a warranty claim are regulated speech, and that holds that line when a shopper pushes for a number.

Frequently asked questions