Lesson 5 of 6

Risky claims and the rules

6 min read

AI writes "clinically proven" with exactly the same confidence as "keeps drinks cold." It has no idea which one it can back up. In marketing, that's not just a mistake — it can be a legal problem.

Confident isn't the same as true

AI predicts words that sound right, so it produces a fluent, sure-sounding line whether or not the fact behind it exists. When it invents something — a study, a statistic, an award — that's a hallucination, and it arrives in the same calm voice as everything true. The tone gives you no warning at all.

AI states a made-up claim in the same confident voice as a true one. Confidence is not evidence — check every claim against something real.

Claims have rules

Advertising isn't a free-for-all. In the US the FTC — the Federal Trade Commission — requires that claims be truthful and backed by evidence; most countries have their own version. "Proven," "#1," health promises, and made-up testimonials are exactly what gets businesses fined. And the rule is simple about blame: you are responsible for what you publish, not the AI that drafted it.

Treat every AI claim as unverified until you've checked it. You own what ships — 'the AI wrote it' is not a defense.

Highest-risk phrases: 'clinically proven', 'guaranteed', '#1', 'cures', and any statistic or testimonial you can't source. If you can't point to real proof, cut the claim or rewrite it.