Privacy policy

Closet AI works almost entirely on your iPhone. Here is what stays on the device, what leaves it, when, to whom and for how long.

Who is responsible

Iván Sevilla, developer of Closet AI (Spain). Contact: ivan.sevilla.ruano@gmail.com.

No account

There is no sign-up or login. We never ask for your email, password or real name.

What stays on your iPhone

• Photos of your clothes and their cut-outs, your wardrobe, your outfits, the calendar of what you wear, tags and where you keep your clothes.
• The name and profile photo you add when you start.
• Your style profile: your answers and what the app learns from your ratings.
• The reference photo for “See it on me” and the avatar made from it. These two files are excluded from iPhone backups.
• The images “See it on me” generates, stored with your clothes.
• Some internal usage counters. They only leave your iPhone if you export the summary from Settings and choose to share it; that summary contains no photos and no storage locations.

The free scan cuts out and classifies the item on your iPhone, using Apple’s tools and our own model, without sending anything. If the item is being worn in a photo, the iPhone detects the silhouette and pose to cut it out better. Nobody is recognised or identified.

What leaves your iPhone, and when

Studio finish (Closet AI Pro)

Only with Pro and the studio finish turned on. When you scan an item, the photo is sent to our service and from there to OpenAI, which separates the item from the background and describes its attributes (type, colour, pattern…).

• If there is a person in the photo, the photo is sent first to locate each item so you can choose which one to save.
• With “Save when it isn’t needed”, a photo on a plain background is cut out on the iPhone, but it is still sent to describe its attributes.
• The photo is downscaled and recompressed before it is sent.
• Without Pro, or with the studio finish off, no clothing photo leaves your iPhone.

“See it on me” (Closet AI Pro)

Only when you tap “See it on me” or “Update on me”, your avatar (your figure cut out from the reference photo, face and body, on a neutral background) and a composition of the outfit’s items, or the image of a single item if you tap it on that item’s page, are sent to our service and from there to OpenAI. OpenAI generates the image of you in that look or item and it comes back to your iPhone. The original reference photo never leaves the iPhone. It is not used to recognise you or to search your photo library.

Identifiers

Every request to our service carries your subscription transaction as signed by Apple. We check it to know the subscription is active, and from it we only store the subscription identifier, which we use to apply usage limits and control cost. It includes the product and its dates, but not your name, your Apple ID or payment details.

It also carries the “identifier for vendor” assigned by iOS, which only appears in the technical logs. It differs for each developer, it is not the advertising identifier, and it changes if you delete all our apps from the iPhone and reinstall them.

Subscriptions

Payments are processed by Apple; we never see your card or payment details. Subscription status is managed by RevenueCat, which receives an anonymous identifier generated by its own SDK, Apple’s purchase receipt and technical device data (device type and system version).

Apple notifies us when a subscription is refunded, revoked or charged again. For a refund or revocation we record that subscription’s identifier, with the reason and the date, so we stop serving it; if it is charged again, we delete the record. The notification does not include your name, your Apple ID or payment details.

If you buy a pack of extra scans, the app sends us that purchase as signed by Apple. We store its identifier, the pack, how many scans it gives, the environment (real or test purchase) and the date, together with your subscription identifier, to keep your balance and never count the same purchase twice. If Apple tells us it was refunded, that pack stops counting. It does not include your name, Apple ID or payment details.

Weather

If you allow location access, the app asks Open-Meteo for the weather using your location rounded to two decimals (just over a kilometre), at most once an hour. The request goes straight from the iPhone to Open-Meteo; we do not receive it. Without permission, the weather is estimated from the time of year.

If you pack for a trip, the app sends Open-Meteo the destination name you type (to find it), then its coordinates and the trip dates (to know the weather there). It goes straight from the iPhone to Open-Meteo; we do not receive it. The trip, its looks and the packing list are stored only on your iPhone.

The morning reminder and the “Today’s look” widget are prepared on your iPhone: there is no notification server and no data leaves the device. A backup is a file you save wherever you choose; we do not receive it.

What our service keeps

• Usage log, in a database in the EU (Frankfurt): for each studio-finish scan, each photo reading and each “See it on me” try-on, the subscription identifier (for uses before 27 September 2026, the device identifier), type of use, model, size and quality, duration, tokens, cost, a request identifier and date and time. It is used to apply usage limits and control cost.
• Server technical logs: the device identifier, type of use, cost, duration, the technical reason if something fails, and the usual data of any web request (IP address, app or browser, path and time).
• We do not keep your photos, your avatar or generated images: they are processed during the request and discarded.

Providers

• OpenAI (USA): processes photos, the avatar and the composition for the studio finish and “See it on me”.
• Vercel (USA): hosts our service.
• Supabase (EU, Frankfurt): hosts the usage log.
• RevenueCat (USA): subscription status.
• Apple: payments, App Store and TestFlight.
• Open-Meteo: weather, with location permission, and the destination of trips you pack for.

Some of these providers process data outside the European Economic Area. They do so as processors, under their data processing agreements and the safeguards those agreements provide for international transfers.

Apart from RevenueCat, the app includes no third-party SDKs: no advertising, no third-party analytics and no cross-app tracking.

How long data is kept

• What is on your iPhone: until you delete it or uninstall the app.
• What is sent to OpenAI: OpenAI does not use it to train its models. It keeps what it receives through its API for up to 30 days for abuse monitoring, unless the law requires longer. An item’s attribute description may also remain in its system for up to 30 days.
• Server technical logs: according to the hosting provider’s plan; currently one hour.
• Usage log: while the service is active, to apply limits. You can ask us to delete it.
• Subscription data: according to RevenueCat’s terms as a processor.

How to delete your data

• An item or an outfit: from its detail screen or its menu.
• Settings → “Delete photo” deletes your reference photo, your avatar and generated try-ons.
• Settings → “Delete the whole wardrobe” deletes items, outfits, the calendar and all images in the app (items, profile photo and generated try-ons). It does not delete your name or style profile, saved places, or the “See it on me” reference photo and avatar; those are deleted with “Delete photo” or by uninstalling the app.
• Uninstalling the app deletes everything it stores on the iPhone (not backups that already exist).
• To have us delete your usage log, write to us. As there is no account, we will help you identify it (for example, by the approximate date of use).

Why we process this data

• To provide the features you ask for, including paid ones (performance of a contract).
• Location and the “See it on me” photo, because you choose to provide them and can withdraw them at any time (consent).
• Usage limits and technical logs, so the service works, is not abused and its cost stays under control (legitimate interest).

Your rights

You can request access, rectification, erasure, objection, restriction and portability by writing to ivan.sevilla.ruano@gmail.com. If you believe they have not been respected, you can complain to the Spanish Data Protection Agency (aepd.es) or your local authority.

What we don’t do

We do not sell or share your data. There is no advertising, no advertising profile and no cross-app tracking.

Children

Closet AI is not intended for children under 14 and does not knowingly collect their data.

Changes

If this policy changes, we will update the date above and let you know in the app about important changes.