AI Transparency Disclosure
Last updated: 2026-05-01
Cosmixxology uses artificial intelligence to analyze hair and scalp images and to recommend personalized formulas. We believe you should understand how the AI works, what it can and cannot do, and what choices you have. This disclosure complements our Privacy Policy and Biometric Notice.
1. What our AI does
- Analyzes photos of your hair and scalp to estimate texture, porosity, density, scalp condition, and color tone;
- Suggests ingredients, recipes, and care routines tailored to those estimates and to the goals you state;
- Generates short text explanations of those recommendations;
- Helps salons match clients to suitable formulas.
2. What our AI does NOT do
- Diagnose disease. The output is cosmetic, not medical.
- Identify you biometrically. We do not run face-recognition, age-estimation, or emotion-recognition models.
- Make legal or similarly significant decisions about you — no credit, employment, insurance, or housing decisions.
- Train on your images. Our processors operate under contracts that prohibit using your inputs and outputs to train their models.
3. AI clearly disclosed
Wherever an output is generated by AI — analysis report, recipe suggestion, summary text — the interface labels it as AI-generated. This is also consistent with FTC guidance on disclosing AI-generated content.
4. How our AI is built
Our AI engine combines proprietary Cosmixxology models, fine-tuned for hair and scalp analysis, with leading third-party vision and language models accessed through enterprise API agreements that prohibit use of your inputs and outputs for model training.
Our infrastructure runs on Microsoft Azure for hosting, storage, and supporting services. We may substitute or supplement underlying providers from time to time as the technology improves. Where the change is material to your privacy, we will update this page and our Privacy Policy.
5. Limitations & known risks
- Results vary with image quality, lighting, and angle. Unusual lighting or motion blur can produce inaccurate estimates.
- The model is trained primarily on adult hair; outputs for children may be less reliable.
- The model performs less well on heavily styled or chemically processed hair where the visible characteristics differ from the underlying biology.
- The model is probabilistic. It will sometimes be confidently wrong. Always sense-check recommendations against your own experience and your professional's judgment.
6. Human oversight
6.1 Built-in safety checks
Before a custom formula is finalized, automated checks run against your declared allergies, the screening rules in Health & Safety, and the ingredient lists permitted in your shipping jurisdiction. Recipes that fail those checks are rejected, not produced.
6.2 Right to human review
You can request human review of any AI-generated analysis or recommendation by emailing support@cosmixxology.com with the subject "Human review". A trained team member will look at the inputs and outputs and either confirm or revise the result.
7. Bias & fairness
We periodically audit model performance across hair textures, colors, and scalp tones to identify and reduce systematic gaps. We publish high-level results with each major model upgrade in our changelog.
8. Your choices
- You can use the marketplace and salon-finder features without ever uploading an image.
- You can delete past analyses any time from Settings.
- You can request human review (see Section 6).
- You can withdraw biometric consent any time (see Biometric Notice).
9. Contact
privacy@cosmixxology.com (privacy/AI questions) · support@cosmixxology.com (request human review)