A sales deck can now be exhibit A. That is the useful lesson from the FTC's finalized Cox Media Group settlement, as reported by Tech Jacks Solutions: the agency did not need a mysterious new AI statute to challenge claims about an AI ad product. It used the familiar consumer protection machinery for a familiar problem, saying the marketing story did not match the product reality. If your AI positioning depends on a capability your engineers cannot substantiate, the compliance risk is no longer theoretical. ## What the FTC actually said, according to All About Advertising Law All About Advertising Law reported that on May 21 the FTC announced proposed settlements with CMG Media Corporation, MindSift LLC, and 1010 Digital Works LLC over an advertising service called "Active Listening." The companies marketed the service to small business customers as an AI-powered tool that could listen to real-time consumer conversations through smart devices, identify when someone was in the market for a product or service, and serve local ads accordingly. According to All About Advertising Law, the FTC said the service did no such thing. That is the regulatory version of asking for the demo logs, not the brochure. Tech Jacks Solutions reported that the FTC later finalized consent orders and that the payment split was $880,000 for CMG, $25,000 for MindSift LLC, and $25,000 for 1010 Digital Works LLC, for a total of $930,000. It also identified CMG Media Corporation as Georgia-based, MindSift LLC as New Hampshire-based, and 1010 Digital Works LLC as Wisconsin-based. The geography is less interesting than the structure: one media company and two marketing firms were all in the frame. The practical point is simple. If several firms touch the claim, several firms may need the evidence file. Passing around the phrase AI-powered does not pass around legal immunity. ## The compliance issue was proof, not microphone folklore Captain Compliance described the case as instructive for AI, consent, and business compliance because it involved a layer of AI hype over claims about consumer conversations and data use. Tech Jacks Solutions reported the FTC's allegation that the service was not based on voice data and that consumers had not opted into the service. That combination matters: the claim concerned both what the system supposedly did and what data it supposedly used to do it. In compliance terms, those are two separate receipts. For builders and marketing teams, the rule is not that every AI claim is forbidden. The rule is that a technical claim needs technical substantiation before it appears in sales copy, pitch materials, landing pages, or partner scripts. If you say a model detects intent from conversations, someone should be able to show how conversation data is collected, authorized, processed, and connected to ad delivery. If the real product uses a different input, the copy needs to say that instead, preferably before the FTC reads it aloud. ## Vendors do not make the claim disappear Andrew Folks at Frankfurt Kurnit framed the FTC's "Active Listening" settlements as lessons on AI marketing, consent, and vendor liability. That is the part many AI resellers and agencies should underline in a color that alarms finance. A vendor relationship may explain how the product was assembled, but it does not automatically cure a deceptive claim made to customers. The customer heard the promise from someone, and regulators are not famous for admiring org charts. All About Advertising Law's account is useful because it shows how the promise was packaged: an AI tool, real-time smart device conversations, in-market identification, and local ad targeting. Translate that into contract work. A reseller agreement for an AI marketing tool should require accurate capability descriptions, documented data sources, consent representations, and a process for approving customer-facing claims. That is not decorative legal furniture. It is how a company avoids discovering, during an investigation, that nobody owned the sentence that sold the product. ## What teams should change before launch, according to Sigma Law Group Sigma Law Group described the $930,000 Cox Media Group settlement as putting deceptive AI marketing claims in the FTC's Section 5 crosshairs. Section 5 is not an AI law, which is precisely why this matters. The FTC does not need Congress to define every model architecture before it can ask whether an ad claim was truthful. Old tools still cut. The operating change is modest but annoying, which is how compliance usually arrives. Before launch, keep a claim ledger that ties each AI capability claim to product evidence, a data map that says what inputs are actually used, a consent position that matches the data flow, and vendor paperwork that mirrors the public story. Legal should not be the last stop after the campaign is polished. If your lawyers say they welcome clarity from regulators, read that as a request for screenshots, model notes, and the current vendor contract. The next cases will not all involve smart speakers or ad targeting. They will involve the same pattern: a product described as more intelligent, more invasive, or more precise than the evidence supports. For builders, the assignment is not to stop describing AI systems. It is to describe them in a way the product team, the data team, and the contract file can survive. ## Sources - The FTC's $930,000 Active Listening Smackdown: What the Cox Media Group Case Means for AI, Consent, and Your Business

Sources