Skip to content
KeepTrip

Privacy Notice

What happens to your photos, where they go, and how long they stay.

Effective date: 2026-10-13

Version 1.8

Draft, version 1.8. A lawyer has not reviewed this text yet, and the highlighted entries below are still to be filled in. It is published so the behaviour of the app is written down somewhere readable, not as legal advice.

The short version

Photos are processed to make your film. If you share it, the film is kept for 7 days, then deleted. No account required.

  • Making and sharing a film needs no account and no email. An optional sign-in by email exists only for someone who wants to keep control of a film later.
  • To build a film, copies of your photos and your notes are sent to AI model providers.
  • If you share a film, it sits at a public web address until it is deleted.
  • KeepTrip's core product sets no cookies and uses no advertising trackers. We count how the app is used, on our own server, with no account and no cookie, and you can turn it off.

1. No account required

KeepTrip has no sign-up, no password and no account you must create to make or share a film.

We do not ask for your name or your phone number, and we do not build a profile of you. If you choose to keep the right to edit or delete a film after the browser that made it is gone, you can give us an email address for that one purpose. Section 7 explains what that involves and how to stop it.

2. What stays on your device

  • The film you are working on, including the resized photos, is held in your browser session storage. Closing the tab ends it.
  • A list of the share links you created is held in your browser local storage, so you can come back later and turn one off. It never leaves the device.
  • Nothing in either store is sent to us on its own.

3. What leaves your device, and when

Nothing about your photos leaves your device until you tap Make the film, apart from the anonymous usage counts in section 9.

The photos themselves leave your device when you ask for a film to be built. At that point your browser resizes each photo to a longest edge of 1024 pixels and re-encodes it as a JPEG. That re-encoding drops the metadata that was embedded in the original file, so the original file itself is never uploaded. A few values read from it are sent separately, as listed next.

What is sent for each photo, in the request that builds the film:

  • the resized copy of the photo
  • the time the photo was taken, read from the original file in your browser before the resize
  • the photo's location and how accurate it is, read from the original file in your browser, so our server can name the place. The server looks the location up in a list of cities that ships with the app and names the nearest city, or the country when no city is close. It does not send the location to a location service.
  • simple camera details such as the lens type, read from the same original metadata

The app does not send your file names, and it adds no details about your browser of its own. Like every website, we receive the ordinary request headers your browser attaches, which include its type. We do send your device's time zone.

Your own context text is sent as well, including any text produced from a voice note.

If you type a place yourself, we send what you typed after removing addresses, coordinates and links; a place you type shows on the shared page.

The exact location is held in memory on our server for the length of that request, then dropped. We do not store it. The AI companies receive the place name, never the location itself or the file names. Text you type, or text printed on a photo, goes to them as text. We remove coordinates and web links from it first, and street addresses where we can recognise them.

Providing photos is not required by law or by contract, but there is no film without them.

When you tap Save video, your browser sends a larger copy of each photo whose original file it still holds, with a longest edge of up to 4096 pixels, unless you choose Save video as is. The video is made from those copies. A photo without one uses the 1024 pixel copy. Your browser re-encodes each larger copy as a JPEG, which drops the metadata embedded in the original file. A copy that fails our check is not kept. We delete the copies when your film is deleted: when you turn the link off, or when the link expires.

4. The AI model providers

The recipients of your content, beyond KeepTrip itself, are Anthropic, Z.ai and Vercel. Two separate AI model roles are used, and they are deliberately different deployments so the second can check the first.

  • The film role writes the captions, the place labels, the trip title and the music pick. It receives the resized photos, the time each photo was taken, the place name, the camera details and your context text. It never receives a photo's location or file name. Provider as currently configured by the operator: Anthropic. If the quality check role cannot be reached, a fallback critic reviews the written text in its place; as currently configured by the operator, that fallback is also Anthropic.
  • The quality check role reviews the written text only. It receives the written film, your title and notes, the time of each photo, the place names, and a short text description of what each photo shows (text on signs, brand and landmark names, a notable detail). It never receives your photos, the file names or the photos' locations. Provider as currently configured by the operator: Z.ai.
  • Vercel, our hosting provider and a US company, runs our servers. Everything described in section 3 passes through Vercel while a film is made. Our server code, the part that handles it, runs in Frankfurt, Germany. Static files and request routing are served from Vercel's network in other places. Vercel also stores a shared film, described in section 5, and keeps request logs (your network address and the page asked for).

No decision with a legal effect or a similarly significant effect on you is made automatically. The AI models write captions, place labels and a music pick; they do not decide anything about you.

Each provider handles what it receives under its own terms, which we do not control. If the operator changes a provider, this notice changes with it. [CONFIRM: transfer mechanism for Anthropic and Z.ai]

5. If you share a film

Sharing writes the resized photos and the text of the film to file storage run by our hosting provider, Vercel, under a long random identifier. The address is public: it is not listed anywhere, but anyone who has the link can open the film without signing in.

The stored copy carries an expiry date set 7 days ahead. After that the share page stops serving the film. An expired link is removed the first time anyone opens it, and a job that runs once a day at 03:00 UTC clears anything nobody tried to open.

If you never share a film, nothing is written to that storage and nothing persists after you close the tab.

The file storage is in Washington, D.C., in the United States (Vercel region iad1), not in the EU. The servers that make the film run in Frankfurt, Germany, so a shared film travels to the United States when it is stored, and opening a link reads it from that storage, possibly through Vercel's content network.

6. Turning a link off

When you create a link, your browser is given a secret token and keeps it locally. Only a one-way hash of that token is stored with the shared film, so the token cannot be read back out of anything we hold.

Turning a link off from the browser that created it deletes the stored photos and text straight away. If that browser loses the token, for example because you cleared site data or switched device, nobody can turn the link off early and it runs out on its own.

For films shared from version 1.26.42 on, copies stop being served within about a minute of turning a link off. Films shared before that can stay in Vercel's content network cache for up to 30 days after they are deleted. A viewer's browser may still hold what it already loaded.

7. Optional: keep control of a film by email

Section 6 lets you turn a link off, but only from the browser that made it. If you want to keep that ability even after that browser, or the site data on it, is gone, you can ask for a one-time sign-in link by email instead.

This is entirely optional. Making, playing, sharing and turning off a film all work exactly the same without it, and we never collect an email unless you explicitly ask for this.

  • We ask for your email only when you choose to keep control of one specific film.
  • We use it only to send you a one-time sign-in link for that film, and to confirm it is you if you later ask us to erase your record.
  • What we keep against your email is a short list of the films you can still control. We never keep a caption, a title, a note or the photos themselves against it.

The sign-in link expires after 10 minutes and works once. Signing in sets a short sign-in cookie on your browser so you can act on the film; it expires on its own soon after, and signing in again always needs a fresh email link.

The legal ground for asking is that you requested it yourself. [CONFIRM: lawful basis for the optional email control feature]

We keep your email and the list of films it controls for as long as at least one of those films is still live, plus 7 more days after the last one expires, in case you want to add another. After that we delete the record automatically.

You can delete your email and this record yourself, at any time, from an active sign-in session. This does not delete the film: it keeps running on its normal 7 day clock and simply stops being something you can edit or delete early, the same state as if you had never asked for this.

The sign-in email is sent through a third-party provider. [CONFIRM: transfer mechanism for the email provider]

8. Voice notes

A voice note is a second way to fill the notes box, not a recording we keep. The speech-to-text conversion is done by your browser. Some browsers do that conversion on servers run by the browser maker, under the privacy terms of that browser. KeepTrip receives only the text that ends up in the box, never an audio file, and stores no recording.

9. Cookies, analytics and logs

KeepTrip's core product sets no cookies. If you sign in through the optional email link in section 7, a short sign-in cookie is set for that purpose only, and it expires on its own soon after. Outside that, there is no advertising code, no third-party tracker and no tracking across other websites. The browser storage described in section 2 is what the app uses to hold your work.

We count how the app is used, so we can see where it breaks and fix it. The counting runs on our own analytics server in the European Union. We do not sell these counts. We do not give them to anyone else. We do not mix them with anything else we hold.

We record which page was opened and what was done there, for example:

  • which screen was opened, with any film or link identifier removed from the address first
  • that photos were added, how many in a rough band such as five to twelve, and roughly what share of them were iPhone HEIC files
  • that a film was requested, whether it was given a name and a note, and how long each step took
  • that a film was played, and whether it was watched to the end
  • that a link was created, copied, opened by someone, or turned off
  • which error appeared, by its code, and whether the device was offline

What is never recorded:

  • the name you give a trip, any caption or your note
  • any file name, any place name, any coordinate, or the time a photo was taken
  • the address of a share link
  • anything read out of a photograph

A count carries only fixed labels, action names from a fixed list, numbers, yes or no answers and fixed codes. It never carries text you typed.

One exception. Before this version of the notice, a mistake in our setup affected counts made after someone opened a shared film. Those counts could include the address of that film and its title. Every count also carried your screen size, your browser language and the page you came from. This version fixes that. Until we finish deleting those older counts, they can still hold all of these.

Each time we count, your browser sends:

  • our website's address and its internal site number
  • which screen you were on, with any film or link code removed, and a short fixed label for that screen
  • for an action, the name of the action and the numbers, yes or no answers and fixed codes listed above
  • a short session code our server gave this tab, so the counts from one visit stay together. The counting code holds it in memory only. It is gone when you close the tab or reload the page.

Nothing else is in a count. The request that carries it also carries what every web request carries: your IP address and your browser's standard details. The next paragraphs say what our server does with those.

The counting code we use can also see your screen size, your browser language and the page you came from. We set it to drop all three, so they are no longer sent.

The counting code saves nothing in your browser. There is no account and no identifier stored on your device for counting. The only thing stored is the switch setting below, and only if you use it. Before each count, the code checks two things: your Do Not Track setting and the switch below. If your browser sends a Do Not Track signal, nothing is counted and the counting code is not even loaded.

Each count is a separate request to our analytics server. From your browser's standard details, the server works out your browser, your operating system and your type of device. From the request's IP address, it looks up your approximate location: country, region and city.

Our counting software does not store your IP address. It turns the IP address and your browser details into a visitor number. So the same browser on the same network counts as one returning visitor. That number changes at least once a month. It is still an identifier. A new number does not delete older counts. We do not attach it to an account or a named person. The older counts described above can still tie a visit to a shared film until they are deleted. Separately, we keep counts for 13 months at most; section 10 says so.

We count without asking you first. Our legal basis is our legitimate interest in understanding and fixing our own product. We have not yet written down the balancing test behind that choice. Our reading is that this needs no consent. One question is still open: whether sending those counts, and holding and sending back that session code, counts as access to your device under German law (section 25 TDDDG). If that question or the balancing test goes against us, we will change how we count and update this notice. [CONFIRM: TDDDG 25 reading of the event sends and in-memory code, and the 6(1)(f) balancing test]

You can turn counting off for this browser with the switch below. It saves one setting in your browser, named umami.disabled, which tells the counting code to send nothing. It applies to this browser only, and you can turn counting back on the same way.

This switch needs JavaScript. Without JavaScript, the counting code does not run either.

Our hosting provider keeps its own standard server logs of requests, which typically include IP addresses, for its own operational and security purposes. [CONFIRM: hosting log retention and terms] The web server in front of our analytics server no longer logs requests to it. Its older logs, which include IP addresses, are deleted by 17 October 2026.

10. How long things are kept

  • Work in progress in your browser: until you close the tab.
  • Your local list of links: until the link expires, you turn it off, or you clear site data.
  • A shared film: 7 days, then removed on first open or by the daily job.
  • An email given to keep control of a film: kept while it still controls at least one live film, plus 7 days, then deleted automatically, or sooner if you delete it yourself.
  • What the AI model providers keep: set by their own terms, not by us.
  • Usage counts on our analytics server: 13 months at most, then deleted.

11. Your rights

If you are in the European Union or the United Kingdom, data protection law gives you rights over personal data about you: to see it, to correct it, to have it deleted, to restrict or object to how it is used, and to complain to your data protection authority. KeepTrip is operated from Thailand, outside the European Union, which can require the operator to name a representative in the EU for GDPR purposes. [CONFIRM: Article 27 representative]

For most of what KeepTrip does, there is no account, so we hold nothing that identifies you and cannot look you up. If you gave us an email under section 7, you can delete that record yourself from an active sign-in session. In practice the fastest routes are the ones in your own hands: turn a link off to delete a shared film, and clear the site data in your browser to remove what is held on your device. For anything else, write to us at hello@jedrzej.com.

12. Children

KeepTrip is not intended for children. You need to be at least 16 years old to use it.

13. Changes, and how to reach us

We can change this notice. The version number and effective date at the top change with it. KeepTrip is operated by Jędrzej Prywata, Chang Klan Road 74/1, 50100 Chiang Mai, Thailand, reachable at hello@jedrzej.com.

The rules for using KeepTrip are in the Terms of Use.