ZetrixwebHow enforcement works
US enforcers have not needed AI-specific statutes to act against exaggerated claims. They rely on existing law. The SEC's first so-called AI-washing cases, in March 2024 against two investment advisers, alleged misleading statements about AI use and ended in settlements. The FTC launched Operation AI Comply in September 2024 and has continued bringing cases under its general deception authority. Law-firm commentary suggests the current FTC has narrowed its focus on AI product capabilities, though settlements continued in 2026 and enforcement in business-to-business contexts has been noted.
Counts of actions vary between trackers, so treat totals as approximate. The consistent point is that misstatements about AI are handled as classic fraud or deception.
Build an evidence file for every claim
For each AI claim, write down exactly what is being claimed, then record how you know it is true. That means the testing method, the data and conditions used, results with limits, and the date. If the claim is about accuracy or savings, keep the underlying numbers. If it is about how the product works, keep an architecture description that matches what actually runs.
Be especially careful with vague, sweeping words such as automated, intelligent, or human-free. If a person reviews outputs or a rules engine does the work, say so or avoid the claim.
| Claim type | Evidence to keep | Common trap |
|---|---|---|
| Uses AI or machine learning | Architecture description and model details | Rules-based system described as AI |
| Accuracy or performance | Test design, data, results, limits | Best-case demo figures |
| Savings or ROI | Customer data and calculation method | Single success story generalized |
| Fully automated | Process map showing human steps | Hidden manual review |
Make it a shared process
Marketing writes claims, engineering knows the truth, and legal knows the risk. Set up a lightweight review where engineering confirms accuracy before publication and the evidence file is linked to the claim. Apply it to the website, sales decks, investor materials, and app store copy.
Revisit claims when the product or model changes, since a true statement can become false after an update.
Need help describing your AI accurately and defensibly? Talk to us about a technical claims review.
Key takeaways
- SEC and FTC actions on exaggerated AI claims rely on existing securities and consumer protection law, not AI-specific statutes.
- Counts of cases differ by tracker; the consistent point is that AI misstatements are treated as deception.
- Keep an evidence file for each claim: what is claimed, how you know, and the limits.
- Set up a shared review between marketing, engineering, and legal, and revisit claims when products change.