Privacy

01Operator

libreAdam is operated by me as an individual, not a company. Every item below is written from the site's actual code; the source notes which file backs which item.

There are no affiliates, so no data is shared with any.

02When you are not logged in

Most toys under /t/ run entirely in your browser. Your input, the images you pick and your progress stay on your machine, are lost when you close the tab or clear browser data, and never reach the server.

The server keeps ordinary access logs: IP, time, requested address, browser UA. Used for debugging and blocking bots, not for profiling.
Language preference is stored in one cookie (la_lang), recording only Chinese or English.
No Google Analytics, and no third-party analytics or telemetry SDK.

03Data stored once you log in

The main site and every subdomain share one account and one table. Sign up once and it works site-wide. The fields are:

Username: Identifies the account. For accounts created with Google, the username is the email address.
Email: Used for logging in, and for contacting you when necessary.
Password hash: Used for password login. Empty for accounts created with Google; the server does not hold your password.
Signup time / last seen: Operational use, to tell whether an account is still in use.
Membership, expiry date, redeemed codes: The basis for quota on paid features.
Usage ledger: One row per successful AI call: feature, month, token count, cost. Neither your input nor the model's output is recorded.
Admin audit trail: Records who changed which account and when, including the operator's own actions.
Job application (if you applied for desk 004): The codename, contact details and answers you filled in. The IP is stored hashed, never in the clear.

04Signing in with Google

The Google button on the login page is rendered by Google's own script, loaded from accounts.google.com. Google therefore knows you opened that page, and may set its own cookies in your browser. That part is governed by Google's privacy policy.

Your browser sends the ID token issued by Google to the server, which verifies its signature and expiry against Google's public keys, and then:

Takes only the email address, plus the single boolean saying whether it is verified.
The name, avatar and Google user ID inside the token are not stored.
No additional scopes are requested; your Gmail, contacts, Drive, Calendar and Photos cannot be read.
After verification it is exchanged for this site's own session; Google's token is not kept.

05Two cookies, total

la_session: Login session, 30 days. It contains one signed username and nothing else. httponly (page scripts cannot read it), https only, SameSite=Lax, scoped to .libreadam.com, so one login on the main site also applies to the subdomain apps.
la_lang: Language preference, Chinese or English.

Deleted on logout. The session is signature-based, so rotating the signing key on the server invalidates every issued session immediately.

No ad cookies, no tracking cookies. There are currently no ads on the site.

06Content you enter

Drafts in the novel workshop, books and vocabulary in the reader, entries and photos in the growth tree, checklists in the moving assistant, and the details entered in the immigration assistant are stored on the server, belong to you, and are visible only to you once logged in.

The growth tree has no public pages and generates no share links.
Not used to train models, not displayed publicly, not sold.
The operator does not read it unprompted, only when you write asking for help with a specific problem, and tells you afterwards what was done.

07EXIF in photos is not stripped

Photos and videos in the growth tree are stored as the original file. The server does not decode images; thumbnails are compressed by your browser and uploaded alongside.

The EXIF inside the file is therefore preserved as-is, including any GPS location the camera wrote. The server reads only the capture time, to sort chronologically, and neither modifies nor removes the rest.

Files are retrievable only once you are logged in and are never public. If you would rather the location data not stay in the file, strip the EXIF before uploading.

08AI features send content to model providers

Features that need a model response send that input, plus the context required, to a model provider. Currently Azure OpenAI (primary), OpenAI and xAI, selected by availability.

The server records only token counts and cost, and keeps no copy of the content. How the provider handles it is governed by their own terms, not by this page.

Do not enter sensitive information such as ID numbers, card numbers or medical records. The same applies to the immigration assistant.

09The data is stored in Germany

The server is a machine rented directly, in Hetzner's Falkenstein datacenter in Germany. The database is a SQLite file on that machine; uploaded photos are files on that machine.

No third-party cloud drive, no managed database, no syncing to anyone else. Apart from the model providers above, the data does not leave that machine.

If you are not in Germany, using the site involves a cross-border data transfer, and using it means you accept that.

10Who can see it

The operator can: the admin console shows the account list, membership status and usage totals, and server logs are read when debugging.
No selling data, no sharing with advertisers, no data exchanges.
There are currently no ads on the site. If ads are added later, this page is updated first.
Except where required by law: a lawful request for disclosure will be complied with, and you will be told as far as that is permitted.

11How long it is kept

Your account and content: kept until you ask for deletion, and not cleared for inactivity.
Server access logs: rotated and overwritten, kept only recently enough for debugging, not archived long term.
Usage ledger: kept, because monthly member quota is computed from it. It contains no content, only counts and token numbers.

12How to delete, how to export

There is not yet a delete-account button on the site. Until there is, this is handled by email: write from the address you signed up with and say whether you want deletion or an export.

Deletion removes the account, the content under it and the uploaded files together, with a confirmation sent afterwards. An export packages your data and sends it back. Handled by one person, usually within a few days.

Self-serve delete and export are planned; this page is updated once they ship.

support@libreadam.com

13Children's information

The growth tree can be used to record a minor. That content is entered by you, and whether you are entitled to record that person is your responsibility.

The rules are the same: no public pages, no model training, nothing to third parties, visible only to you once logged in.

This site is not intended for children under 13 to sign up and use on their own.

14Your rights

You can view, correct, export and delete your own data by asking over email.

If you are in Europe (GDPR) or Canada (PIPEDA), the rights those laws grant are the ones above, handled as described. You may also complain to your local regulator at any time.