Trust and safety

Cosmetic, not medical: where an AI skin analyzer has to stop

An AI skin analyzer for beauty brands should describe what is visible and stay out of diagnosis. How the cosmetic line is drawn, enforced and explained to customers.

13 November 2026 · 7 min read · By the Claré AI team

Skin is personal, and people read a result as a verdict. Show someone a number next to the word skin and they will take it seriously, sometimes more seriously than the number deserves. That is why the boundary between a cosmetic tool and a medical one has to be a product rule, not a footnote at the bottom of a page.

What cosmetic analysis means in practice

Cosmetic analysis looks at visible, everyday skin concerns and suggests care products. Claré covers seven: blemishes, dark spots, dark circles, oiliness, dryness, redness and fine lines. It reports how strong each one looks, mild, moderate or severe, and where on the face it was seen. It does not diagnose, does not detect disease and does not use medical terminology.

The distinction matters because the same visible mark can mean very different things. A flat dark mark may be harmless pigment left after a spot has healed, or it may be something a doctor should look at. A camera app cannot tell the difference, and it should not pretend to. The general background on how skin is structured and what can affect it is well covered in this overview of human skin, which also shows how much a photo cannot capture.

Why a disclaimer alone is not enough

Many tools add a small line saying this is not medical advice and then go on to use clinical words in their results. Customers read the results, not the footer. If the output says something that sounds like a diagnosis, the disclaimer will not rescue the customer experience, and it will not protect the brand either.

Enforced in two places

  • In the instructions the AI receives, which keep it to cosmetic language and everyday concerns.
  • In a validator that screens the output afterwards, so a slip at the first stage does not reach the customer.

Both layers exist because a single layer eventually fails. Models drift, prompts get edited and edge cases appear. A second check turns a policy into a property of the system.

Language that stays on the right side of the line

Good cosmetic copy describes what is visible and offers care. It avoids naming conditions, promising outcomes or telling someone what is wrong with them. Compare these two sentences. The first, there are flat marks on both cheeks, describes. The second, you have a pigmentation disorder, diagnoses. Only the first belongs in a beauty product.

It also means being careful with age. Claré never estimates age from a face. Age is an optional dropdown that the customer can skip, and it changes the wording of a result, never the score. We explain why in our post on selfie skin scans and privacy.

What to tell your customers

Be plain about it, next to the camera button and again on the report. Something like: this is a cosmetic analysis to help you choose products, it is not medical advice. Every Claré report says so in the footer, and your own disclaimer text is shown at the start of the flow, in your colours.

Why this is good for brands too

A beauty brand that stays on the cosmetic side of the line keeps its tone honest and its risk low. Shoppers still get something useful: a clear picture of what they might want to care for and a routine that fits it. The brand gets a tool it can put in front of customers without a legal conversation every time the model changes.

Regulators in different countries treat health claims differently, so if you operate across markets, check the rules where you sell. Our own cosmetic disclaimer and terms are drafts that your counsel should review for your situation.

When to point people elsewhere

If someone has a concern about their skin health, such as a changing mole, persistent pain, bleeding or an unusual rash, a cosmetic scan is not the place to resolve it. A good tool says so in the product, in the report and on the website. We do all three.

A short checklist for cosmetic wording

If you write copy around a skin tool, or review what a vendor’s tool says to your customers, a few habits keep you safe and keep the tone friendly. None of them are complicated, but they are easy to forget when a launch deadline is close.

  • Describe what is visible, in everyday words. Flat marks on both cheeks, a little redness around the nose, mild darkness under the eyes.
  • Offer care, not cures. Talk about routines, protection and gentleness, never treatment or results you cannot guarantee.
  • Avoid names of conditions. If a word would appear on a medical form, it does not belong in a shopping flow.
  • Say what the tool cannot know. One photo cannot see everything, and a good result admits it.
  • Keep the disclaimer visible at the start and the end, not only in the footer of the site.

These habits also make your customer service easier. When a shopper writes in worried about a result, you can answer calmly: it is a cosmetic description meant to help you choose products, and if you are concerned about your skin, a doctor is the right person to ask. That answer is honest, short and kind.

If you want to see how this plays out in a live flow, read how Claré works or ask us for a demo on a sample catalog.

Frequently asked questions

Does AI skin analysis diagnose skin conditions?

A cosmetic analysis does not. It describes visible, everyday concerns and suggests care products. Claré never diagnoses or uses medical terminology.

Where does Claré say it is not medical advice?

In your disclaimer text at the start of the flow, in the analysis wording itself and in the footer of every PDF report.

Cosmetic analysis only. Not medical advice.

Ready when you are

See Claré on your catalog.

Book a demo and we will walk through a scan, the checks and a routine built from a sample catalog, then talk about how it fits your brand.