AI Policy & Transparency Notice
Companion to the Privacy Notice, Trust Centre, and Acceptable Use Policy. Reflects our obligations under the EU AI Act (transparency obligations apply from 2 August 2026) and UK law.
How Onyou uses AI
Onyou is an AI styling tool. When you use it, AI reads your selfie to work out your color season and coloring, and generates styled looks for you. This notice explains — plainly — what the AI does, what it doesn't do, and the rules we hold it to.
You're always working with AI — and we label it
- The analysis and looks are AI-generated. Onyou is an AI system; you're never talking to a person pretending to be a bot, or the reverse.
- Every look image is clearly labeled "AI-generated." These are AI mock-ups and a styling opinion — not real photographs, and not a promise of a result. We keep that label on the images, and we ask you to keep it if you share them.
What our AI does
- Reads your coloring — skin undertone, hair, eyes, contrast — to place you in a color season and suggest palettes.
- Generates styled looks — hair, clothing, jewelry, and makeup that suit your coloring.
- Applies subtle, natural makeup. Your free reveal includes light makeup styled to your season: just subtle, natural makeup — your own features, unchanged. No reshaping, no filters. We do not enlarge eyes, slim jaws, or smooth skin. Styling changes what you're wearing, not who you are.
- You steer it. You choose the vibe from set options; the AI works within the styles we've designed, not open-ended instructions.
- It's an opinion, not a verdict. Results are styling guidance and entertainment. They don't decide anything about you, and you're free to ignore them.
What our AI does not do
- It doesn't identify you. We scan an upload only to confirm there's a face in it — never to work out whose face it is. No facial recognition, no matching you against any database.
- It doesn't categorize you by race, ethnicity, or any protected characteristic. We analyze visible coloring for styling purposes only. We don't infer or label your race, ethnicity, health, beliefs, or any sensitive trait, and we don't store any such label.
- It doesn't read your emotions.
- It doesn't make automated decisions that have legal or similarly significant effects on you.
- It isn't a deepfake, face-swap, or impersonation tool. It styles you, from your own selfie. (Full detail in the Trust Centre.)
- It doesn't "improve" your face. No beautification, no reshaping, no filters — identity is held, deliberately.
Fairness
We tested Onyou across a wide range of skin tones and ethnicities — including deeper skin tones and Indian and East Asian faces — specifically to hold your identity and coloring as faithfully as we can. Looking like you is the point, and it's something we actively test for. AI generation isn't perfect and a likeness can vary, so we re-run generation while we're creating your look if the likeness drifts. (Once your looks are made, we don't recreate them — see the Privacy Notice.)
The AI models behind Onyou
We use trusted, enterprise third-party AI models, each under a data-processing agreement:
- An AI vision model reads your selfie and writes your analysis.
- AI image-generation models create your looks.
- An independent safety scanner checks every uploaded and generated image for nudity and violent content and blocks anything it flags.
Your photo creates your look — then it's gone. We never keep it, never use it to train AI, and handle it under formal data-protection agreements.
Human oversight and your control
- You consent before any photo is processed, with a separate, un-ticked checkbox.
- You steer the styling from set options.
- You can withdraw consent and delete anything we've stored, any time.
- Because results are a styling opinion, the human in charge is always you.
Our position under the EU AI Act
We treat Onyou as a limited-risk AI system subject to transparency obligations — not a high-risk system. We meet those obligations by telling you clearly that you're using AI and by labeling AI-generated images. We do not operate any AI practice the Act prohibits: no untargeted facial-recognition scraping, no social scoring, no emotion recognition, and no biometric categorization to infer sensitive characteristics.
Questions & concerns
Questions about how our AI works, or want to raise a concern? Email support@onyou.beauty (or privacy@onyou.beauty for anything about your personal data).
Changes
We'll update this notice as the product and the law evolve, and update the date above.