The Federal Trade Commission will require Cox Media Group and two marketing partners to pay a combined $930,000 to settle charges that their "Active Listening" product — pitched as an AI system that targeted ads built from real-time voice conversations captured by phones and smart speakers — did no listening at all. The order, approved on a 2-0 Commission vote, is the agency's clearest move yet against AI-washing: selling a conventional service under an artificial-intelligence label.
What was sold vs. what shipped
Cox Media Group (CMG), MindSift, and 1010 Digital Works marketed Active Listening as proprietary technology that detected relevant conversations overheard by smart devices and converted them into geo-targeted ad audiences. According to the FTC, none of that happened. There was no voice capture, no on-device audio model, no real-time speech analysis. The actual product was email lists purchased from third-party data brokers, resold at a significant markup — and even the promised geographic targeting was inaccurate.
CMG carries the bulk of the penalty at $880,000; MindSift and 1010 Digital Works pay $25,000 each. The CMG portion is set aside to refund customers who bought the service.
The consent finding builders should read twice
The second prong of the complaint has the most direct relevance to anyone shipping data products. The companies told clients that consumers had "opted in" to Active Listening because those consumers had accepted standard app terms of service. The FTC rejected that outright: clicking through mandatory ToS is not opt-in consent for voice-data collection from inside someone's home. "It is a basic rule of business that you need to be honest with your customers, and these companies failed to do that," said Christopher Mufarrige, director of the FTC's Bureau of Consumer Protection.
The consent orders bar all three firms from misrepresenting their products' features, any collection of voice data or consumer consent practices, and geographic-targeting capabilities.
Why a fake AI product is an AI-policy signal
The irony is that Active Listening never used AI — and that is precisely the enforcement hook. The FTC is treating the claim of AI capability as a material representation that has to be true. For vendors shipping genuinely AI-driven ad, analytics, or personalization tooling, the order draws a hard line: capability claims ("our model analyzes…", "real-time on-device inference") are actionable if the system doesn't actually do them, and consent harvested through buried ToS language won't survive review.
For enterprise buyers, the takeaway is procurement-grade: diligence the data pipeline behind any "AI-powered audience" pitch, demand documentation of what the model actually ingests, and treat opt-in claims as something to verify rather than assume. Reselling broker lists with an AI wrapper was always a thin product. It is now also a $930,000 liability — and a template for how the FTC will police the next vendor that overstates what its model does.
