English

This page explains exactly what is — and is not — sent to our third-party AI providers. There are two separate uses, with two providers.

1. Reading product labels

When you take a photo of a product, that photo is sent to Anthropic only to read the printed text: the ingredient list, the product name, the brand and the kind of product. The screen tells you before the first photo, and nothing is photographed until you tap the shutter. We don't store the photo: it is processed for that one request and discarded. Please point the camera at the product, not at people.

The AI does not judge products. Scores and verdicts are calculated on our own servers, by fixed rules, from the ingredient text and your skin profile. Searching by name or pasting an ingredient list sends nothing to Anthropic.

2. Optional AI explanations

SkinwellAI can write an optional plain-language explanation of the estimates from a photo skin profile. Your face photo is never sent to a third-party AI: image analysis (MediaPipe/OpenCV) happens on our own backend. For this explanation we send OpenAI only:

Consent is required and reversible. These explanations run only after you explicitly opt in. If you decline, the text is generated on our own servers with no third-party AI. You can grant or withdraw consent any time in Profile → Privacy & data; if our policy version changes, we ask again.

Safety. Generated text is filtered for non-diagnostic, non-shaming wording and validated against your real scores; if it cannot be validated, we fall back to a deterministic, on-server text.

Provider terms

Under Anthropic's and OpenAI's commercial API terms, inputs are not used to train their models by default, and we do not use this data for advertising.