Claims you can't make — and how to say it instead
K-beauty copy translates badly into US captions. A phrase that is ordinary on a Korean product page can read, in English, as a claim that the product is a drug. You post directly to TikTok and nothing is reviewed first, so it is worth knowing the words to leave out — and how to describe the same result without them.
The line the words sit on
A cosmetic is sold to clean the body or change how it looks. Wording that says a product treats or prevents a condition, or changes how the body works, is the wording used for drugs — a different category with a different set of requirements.
Which side a product is sold on is decided largely by the language around it. That is why a caption can create a problem the product itself does not have.
The registration side is the brand's, and it is settled before the campaign reaches you — a brand cannot create a campaign until its compliance record is verified. What is left on your side is what you say.
The terms to leave out
This is the list we give every creator. Capitalisation does not matter for any of these.
| Group | Terms to avoid |
|---|---|
| Drug-like claims | cure · cures · cured · heal · heals · healed · treatment · prescription · FDA approved · FDA-approved · clinically proven · permanent results · permanently fixes |
| Condition claims | treat · treats · curing · healing — when followed by acne, eczema, psoriasis, rosacea, dermatitis or wrinkles |
| Side-effect claims | no side effect · no side effects · zero side effect · zero side effects |
| Superlatives | #1 (before a word) · number one · world's best · best in the world · miracle |
There is no allowlist on the other side. A word simply not being on this list is not the same as it being safe — the last section is about that difference.
Three that are easy to miss
treatment
A problem even as a product category. “Spot treatment”, “treatment essence” and “acne treatment” all say the product treats something, even though that is how it is sold elsewhere. Describe the step instead of naming it.
FDA approved
The FDA does not approve cosmetics the way it approves drugs, so the phrase does not apply to one. Hyphenated or not.
clinically proven
Leave it out whether or not a study exists. If the brand has data behind a claim, that claim belongs on the brand's own listing — it does not transfer into your caption.
Say it as what you saw
Every one of those phrases has a version that describes your own result instead of the product's power.
| Instead of | Write |
|---|---|
| cures acne | my breakouts calmed down over three weeks |
| heals the skin barrier | my skin stopped feeling tight |
| acne treatment | the step I use on breakouts |
| clinically proven to fade dark spots | my dark spots look lighter than they did in June |
| no side effects | it didn't sting, and I have reactive skin |
| the #1 serum in Korea | the serum I've repurchased three times |
| permanent results | still holding up three months in |
| miracle product | the only one that stayed in my routine |
The pattern underneath all of them: first person, what you observed, over a period you name.
What the list can't do
It covers captions and speech alike. Anything you say out loud or put on screen is as much a claim as anything you type — and TikTok applies its own rules to all of it. A claim you would never type is still a claim when you say it.
It is not a complete list either. A drug-like claim written in words that are not on it is still a problem when a person reads it. Rewording to slide past a list is not what any of this is for — the list exists because those words turn a description of a cosmetic into a claim about what it does to your body.
These guides describe how Welike campaigns work as of the date above and, in general terms, what US advertising rules require of a post. They are not legal advice and not a complete account of US cosmetics regulation. If a brief asks you to say something you are unsure about, ask before you post.
Read next
Still unsure about a line in a brief? Ask before you post — not after.
