The FTC Just Fined an “AI That Listens to Your Conversations” — It Never Existed

What Happened: The $930,000 “Active Listening” Settlement

On August 27, 2026, the U.S. Federal Trade Commission finalized consent orders against Cox Media Group (CMG) and two marketing firms, MindSift LLC and 1010 Digital Works LLC, over a product called “Active Listening.” The service was pitched to small businesses as an artificial intelligence system that could detect purchase intent from conversations captured by consumers’ smartphones, smart TVs, and smart speakers — then target ads to those people within a defined local radius. Customers were also told that consumers had “opted in” to this voice-based ad targeting.

The penalties break down as $880,000 from Cox Media Group and $25,000 each from MindSift and 1010 Digital Works, totaling $930,000, which will fund redress for affected CMG customers. The Commission voted 2-0 to approve the final orders after a public comment period. The orders also impose 20 years of reporting and recordkeeping obligations and prohibit the companies from misrepresenting their advertising services, their collection and use of voice data, consumer consent claims, and geographic targeting capabilities.

The Twist: The AI Never Existed

Here’s the part that should reshape how you evaluate every AI-branded tool you’re offered: according to the FTC’s complaints, the service was not based on voice data at all. What customers actually received were consumer email lists purchased from third-party data brokers and resold at a markup — with geographic coverage that bore little relation to the territories businesses thought they had paid to reach.

The companies had leaned on consumers accepting app terms of service as evidence of “opt-in” consent. The FTC rejected that reasoning outright, noting that accepting mandatory app terms is not consent for an invasive service involving voice data from inside people’s homes. The agency added a remarkable footnote: had Active Listening actually worked as advertised — collecting and using voice data without adequate consent — the practice itself would have violated Section 5 of the FTC Act. In other words, the only thing worse than the lie would have been the truth.

The conduct at issue ran from 2023 to mid-2024, and the case traces back to the FTC’s Operation AI Comply, the enforcement sweep launched in September 2024 against companies exploiting artificial intelligence claims.

A Pattern: 14 Enforcement Actions, $51 Million Recovered

If you think this is a one-off, it isn’t. To date, every AI-related enforcement action the FTC has brought has targeted marketing deception rather than the underlying AI behavior itself. Under the Operation AI Comply umbrella, the commission has now filed 14 such actions and recovered nearly $51 million in total. The regulatory focus is consistently the gap between what a company promises its AI can do and what the technology actually delivers.

This matters for the entire AI tools market. The industry has a term for it — AI washing — and it ranges from sloppy exaggeration (a rules-based script marketed as “AI-powered”) to outright fabrication, as in the Active Listening case. When the headline capability of a product is the AI itself, and the AI doesn’t exist, buyers aren’t just overpaying — they’re making business decisions on fiction. We covered how to spot the milder forms of this in our guide to AI washing and fake AI tools, and the FTC’s latest action raises the stakes considerably: fake AI claims now carry binding, decade-long consent orders, not just refunds.

How to Spot Fake AI Claims Before You Pay

Small businesses paid real money for a phantom capability. A short due-diligence checklist protects you from the same trap:

• Ask what data the AI actually processes. If a vendor can’t explain the input, the model, and the output in plain language, treat the “AI” label as decoration. In the CMG case, a single direct question about the voice pipeline would have collapsed the story.
• Demand a measurable capability, not a vibe. “Detects purchase intent from conversations” is a testable claim. Ask for a trial, a methodology doc, or third-party verification before signing.
• Be skeptical of surveillance-flavored superpowers. Claims that a tool monitors private conversations, reads minds, or taps data no lawful product could access are either illegal or false. Usually false.
• Check the consent story. “Consumers opted in via app terms of service” is exactly the argument the FTC just rejected. If consent sounds like a legal fiction, the capability probably is one too.
• Google the vendor + ‘FTC’ or ‘complaint’. Operation AI Comply actions are public, and 20-year consent orders are discoverable in seconds.

What Real AI Marketing Tools Actually Do

The ironic backdrop to all this: genuinely useful, honestly marketed AI marketing tools have never been more capable — and none of them need to eavesdrop on your living room to deliver value. A few examples from our directory that do what they say on the label:

Notice what these have in common: the capability is demonstrable in a free trial, the data inputs are obvious, and the claims are boring in the best way. Boring claims are how you know it’s real.

Frequently Asked Questions