AI Washing: AI Claims That Get You in Trouble
‘AI-powered’ is on almost every SaaS homepage. When the claim is vague, overstated or untrue, it stops being marketing and becomes a legal and trust problem. Regulators have said clearly that there is no AI exception to advertising law.
What is AI washing?
AI washing is overstating what AI does in a product: calling a feature AI when it isn’t, claiming results the AI can’t deliver, or implying the product replaces experts when it doesn’t. In September 2024 the FTC announced five actions under Operation AI Comply, and its chair said there is no AI exemption from the laws on the books (FTC on Operation AI Comply).
What did regulators object to?
The FTC’s 2024 sweep included a company marketing ‘the world’s first robot lawyer’ and several schemes promising AI-powered income. The lesson for SaaS is not that AI claims are banned, but that they must be true, specific and supported.
| Risky claim | Why it’s risky | Safer version |
|---|---|---|
| ‘Replaces your finance team’ | Implies expert substitution it can’t deliver | ‘Drafts invoice approvals for your team to review’ |
| ‘100% accurate AI’ | No AI system is error-free | ‘Extracts invoice fields; you confirm before posting’ |
| ‘AI-powered’ on a rules engine | It isn’t AI | Describe the automation plainly |
| ‘Saves 10 hours a week’ with no data | Unsupported result claim | Quote a measured result with its source, or don’t quote one |
What should you check before publishing an AI claim?
Before publishing an AI claim, check three things: does the feature actually use AI in the way the claim suggests, can you support every result or comparison with evidence, and have you said what it can’t do or where a human must check. Keep the evidence on file.
- Does it really use AI as described?
- Can we prove every number and comparison?
- Have we stated limits and where people must review?
- Would a buyer feel misled after using it for a week?
How should you describe AI features instead?
Describe the job and the result, then the role of AI. ‘Drafts a reply from the ticket history; you edit and send’ tells buyers exactly what happens. More on outcome-led messaging is on positioning AI features, and on the honest UX that backs it up on designing AI users can trust.
Our revolutionary AI replaces manual work forever with 100% accuracy.
Ledgerly reads each invoice PDF, fills in the fields and flags anything unusual. Your team reviews and approves.
The honest version also converts well with careful B2B buyers, because it answers the question they actually have: what will this do in my team’s day?
Specific beats spectacular.
A precise claim you can prove builds more trust, and more sales, than a big claim buyers learn to distrust.
Who should review AI claims?
Have product, marketing and someone responsible for legal review sign off on AI claims on the homepage, ads, sales decks and launch posts. Re-check claims when the model or feature changes. Outside the U.S., other rules apply too; the EU AI Act adds transparency duties for some AI uses, covered on AI, data and security. Launch messaging is on launching an AI feature. This is general information, not legal advice.
Frequently asked questions
What is AI washing?
Overstating what AI does in a product, such as calling non-AI features AI or claiming results it can’t deliver.
Is AI washing illegal?
Misleading claims can break advertising and consumer protection law. The FTC has said there is no AI exemption.
What was Operation AI Comply?
An FTC sweep announced in September 2024 with five actions against deceptive AI claims and schemes.
Can I say my product is AI-powered?
Only if it is, and it’s better to say exactly what the AI does and where people review its work.
Do I need evidence for AI result claims?
Yes. Support any time saved, accuracy or comparison claim with evidence you keep on file.
Who should approve AI marketing claims?
Product, marketing and someone responsible for legal review, re-checked when the feature changes.
Sources & further reading
Standards and platform rules change. These sources let you verify the current requirements directly. All screens shown are mock-ups of fictional products.