How to Market AI Without Crossing the FTC's Line

There is now a name for dressing up an ordinary product with artificial-intelligence language: AI washing. And there is a federal enforcement program built around it.

Since September 2024, the Federal Trade Commission has been running a sweep called Operation AI Comply, suing companies that either claim AI does things it cannot do, or use the word "AI" to make a weak or fraudulent offer sound cutting-edge. The cases have continued under new agency leadership, with fresh settlements landing through early 2026.

For most AIRegReady readers, this is not an abstract Washington story. Two very ordinary situations put a business in the FTC's path: marketing your own product as more "AI-powered" or automated than it really is, and buying into a "make money with AI" program that cannot deliver. Both are exactly what the FTC has been targeting.

The rule the agency keeps repeating is short. As then-Chair Lina Khan put it when the sweep launched: "There is no AI exemption from the laws on the books."

What "AI Washing" Actually Means

AI washing is the marketing cousin of greenwashing. It covers a few related habits:

• ●Claiming AI that barely exists. Calling a product "AI-powered" when the real work is done by simple rules, templates, offshore labor, or the customer.
• ●Overstating what the AI can do. Promising accuracy, automation, or results the system has never been tested to deliver.
• ●Using "AI" to sell an outcome. Wrapping a get-rich-quick or business-opportunity pitch in AI language so it sounds modern and inevitable.
• ●Selling tools that produce deception. Offering AI that generates fake reviews, fake testimonials, or other content designed to mislead other people.

Operation AI Comply: The FTC's AI Marketing Sweep

The FTC announced Operation AI Comply on September 25, 2024, with five enforcement actions filed at once. They fell into recognizable patterns that any founder or operator should be able to spot:

• ●DoNotPay marketed an "AI lawyer" — a "robot lawyer" that could "generate perfectly valid legal documents in no time." The FTC said the company employed no attorneys and never tested its chatbot against a real lawyer. A finalized 2025 order required DoNotPay to pay $193,000 and barred claims that its tools can replace licensed legal professionals.
• ●Ascend Ecom, Ecommerce Empire Builders, and FBA Machine were three separate "passive income" schemes that leaned on AI credibility to promise effortless online-store profits that customers did not actually earn.
• ●Rytr sold an AI writing assistant that the FTC said could churn out fake, detailed reviews — giving users "the means and instrumentalities" to deceive consumers at scale.

The Crackdown Did Not Stop With New Leadership

A fair question in 2026 is whether this is still a priority under an administration that talks openly about reducing AI regulation. On deceptive marketing, the answer so far is yes. The cases kept coming:

• ●Click Profit (March 2025) — Promised AI-run passive income. The FTC said about 20% of customers earned nothing and another third earned under $2,500. The action produced more than $20 million in judgments.
• ●Workado (April 2025) — Advertised an "AI Content Detector" as 98% accurate when independent testing put real-world accuracy closer to 53%. No monetary penalty, but the company was ordered to stop the accuracy claim and submit to monitoring.
• ●Air AI (settled March 2026) — Marketed a conversational AI that could "replace" human customer-service reps and, bundled with other services, make owners millions. The FTC said some buyers lost as much as $250,000. The settlement carried an $18 million judgment (largely suspended for inability to pay, with $50,000 paid for consumer relief), banned the operators from marketing business opportunities, and barred earnings claims without real substantiation.
• ●Growth Cave (settled January 27, 2026) — Claimed its "GrowthBox" software would "automate nearly 100% of the process" of setting up online courses, when users still had to do the work manually. The order prohibits misrepresenting that a product uses AI to boost profitability, effectiveness, or efficiency.

One Case the FTC Walked Back — and Why It Matters

The picture is not "the FTC will sue anything with AI in it." In December 2025, the agency reopened and set aside its own 2024 order against Rytr, concluding that the original complaint did not meet the FTC Act's requirements and that the order "unduly burdens artificial intelligence innovation." The agency framed it as the first action implementing the current administration's AI Action Plan.

Read the two trends together and the practical lesson gets sharper. The FTC has shown reluctance to punish a company simply for *building an AI tool* that someone else might misuse. But it has stayed aggressive against businesses that make false earnings, performance, or capability claims — whether or not AI is involved. The deception is the target. AI is just the costume it often wears now.

The Two Ways Ordinary Businesses Get Caught

You do not need to be a frontier AI lab to land in this enforcement lane. The risk shows up in two everyday roles:

As a seller. The moment you put "AI-powered," "fully automated," "10x your output," or a specific accuracy number on a page, you have made a claim you may have to prove. AI tools make it easy to generate confident marketing copy fast — which is exactly how unsupported claims end up published before anyone checks them.

As a buyer. The "earn passive income with AI," "replace your staff with AI," and "automate 100% of X" pitches are not just annoying. They are the precise category the FTC has been dismantling. If a program guarantees results, leans on AI mystique, and gets vague about how the money actually appears, treat that as a red flag, not a strategy.

Claims That Draw FTC Attention

A few phrasing habits reliably turn a marketing line into a substantiation problem. Watch for these in your own copy, sales decks, ads, and chatbot scripts:

• ●"AI-powered" or "AI-driven" for something that is mostly rules, templates, manual work, or a thin wrapper around someone else's model.
• ●"Fully automated" or "automates 100%" when a human still does meaningful steps.
• ●Specific accuracy or performance stats ("98% accurate," "saves 20 hours a week") with no testing behind them.
• ●Earnings and results promises ("make $10,000 in 30 days," "guaranteed refund") tied to an AI product.
• ●"Replaces a lawyer / accountant / employee / agent" when the tool has never been validated against that work.
• ●AI-generated reviews or testimonials, or tools sold mainly to produce them.

How to Market AI Without Crossing the Line

You do not have to stop saying you use AI. You have to be able to back up what you say. A short discipline keeps most businesses clear:

• ●Describe what the AI actually does. "Drafts a first version you edit" is safer and more honest than "automates your whole workflow."
• ●Substantiate performance claims before publishing. If you cannot point to a test, a measurement, or a source, soften the claim or drop it.
• ●Avoid earnings and outcome guarantees. Results that depend on the customer's effort, market, or skill are not yours to promise.
• ●Disclose human involvement honestly. If people review, edit, or do part of the work, do not market it as untouched automation.
• ●Never sell or generate fake reviews. This was a named target in the very first sweep.
• ●Keep a claim-evidence record. For each meaningful AI claim, note the exact wording, who approved it, the evidence behind it, and when it was last checked. (We walk through this file in the post on owning AI-generated claims.)

If You Are Buying AI Tools or Programs

The same enforcement record is a buyer's checklist in disguise. Before you pay for an AI tool, course, or "business opportunity," apply a little friction:

• ●Be skeptical of guaranteed money. Legitimate tools sell capability, not promised income.
• ●Ask what the AI literally does versus what the marketing implies. Make them describe the workflow in plain terms.
• ●Test the headline claim when you can. A free trial or a small pilot will expose a "98% accurate" promise quickly.
• ●Read the refund terms, especially when a refund "guarantee" is the main reason the offer feels safe.
• ●Treat "automates 100%" as a question, not a feature. Find the manual steps before you commit a budget or a customer workflow to it.

The Bottom Line

AI has made it cheap to produce confident, polished, technical-sounding claims. It has not lowered the bar for proving them. That is the whole message of Operation AI Comply, and it has survived a change in administration: the agency softened on punishing AI tool-makers, but it is still going after deceptive earnings, performance, and capability claims.

For a small business, the takeaway is manageable. Say what your AI actually does. Keep evidence for the claims you publish. Be the skeptical buyer when someone promises AI riches. None of that requires a compliance department — just the discipline to make sure the marketing and the reality describe the same product.

Related Regulations

Sources & References