Analytics choice

Google Analytics may send limited cookieless measurements. Accept enables analytics cookies; advertising tracking remains disabled. Details

Skip to main content
Jury
Jury AIJury GraphLaw libraryFor teamsPlatformGet startedFAQ
МНEN
Open chat
Open chatJury AIJury GraphLaw libraryFor teamsPlatformGet startedFAQLet’s talk
LEGAL

Privacy Policy

This policy explains what information jury.mn and chat.jury.mn process, why it is used, and who may receive it.

Last updated: 12 November 2026
Before you submit information

Jury AI is operated by Смарт интел ХХК, registration number 6986897. The service may send information to foreign providers when processing answers and searches. Do not submit sensitive, confidential, professionally privileged, or third-party personal data unless necessary and lawful. Redact names, identification numbers, phone numbers, addresses, and other identifiers. See section 4 of the Privacy Policy for details about providers.

1. Service operator

jury.mn and chat.jury.mn are operated by Смарт интел ХХК, registration number 6986897. Address: 11th khoroo, Sukhbaatar District, Ulaanbaatar 14190, Mongolia. Phone: 88110921. Email: info@jury.mn.

2. Information we process

We process the following information as part of operating the service. You choose what to include in a question, and legal questions may contain sensitive information.

  • your verified phone number if you sign in with one; account and login timestamps; and authentication-token records;
  • if you sign in with Google or Facebook, the provider’s stable user identifier, the time it was linked to your Jury account, and the name and email address the provider shares (for Google, only a verified email). The name and email are used only to show which account you are signed in to and for support, never to find or merge accounts. A Google profile-picture URL is not saved;
  • if you sign in through your organization (single sign-on), the address of the identity provider your organization configured, your stable identifier there, the verified email address it shares, and a record of the sign-in in your organization’s audit log;
  • if you turn on two-step sign-in, your authenticator secret in encrypted form and your recovery codes only as one-way hashes;
  • questions, answers, conversation titles and history, and saved legal sources;
  • PDF, Word and text documents and images you attach, their filenames and sizes, and text extracted from documents;
  • information about conversations, models, tokens, costs, usage allowances, and usage totals;
  • product code, amount, status, transaction identifiers, provider, timestamps, and payment-gateway event information;
  • organization name, registration number, billing email, membership, invitation, role, shared research matters and their access log, and audit information;
  • if you use an organization’s channel (chat.jury.mn/c/…), that you joined it and when; if the channel admits people from a list, whether your phone number or email matches that organization’s list; if you send your case to an organization, the summary you checked and agreed to, the information you selected, the conversation transcript if you chose to include it, the phone number or email you gave, the time and version of your consent, and the messages you then exchange with the organization; if you agree to topic statistics, the general topic of your questions in the current month;
  • IP address and request, security, and error-related technical information; language, authentication, analytics, and ad-measurement choices stored in the browser.

3. Why we use it

We use information to authenticate users, answer questions, search legal sources, operate conversation history and saved sources, track usage allowances, confirm payments, operate team features, send notices, protect the service, investigate abuse or errors, and respond to requests.

When you use an organization’s channel, information is used to admit you to the channel, to work out whether the channel pays for your questions, and to pass your case to the organization you chose. A case is passed on only when you agree and send it, each time.

We process information to provide the service, perform our relationship with users, meet legal obligations, protect service security, or, where required, based on your consent. We request separate consent for foreign transfers or sensitive information when required by law.

You can sign in with a phone number, Google or Facebook, or through your organization if it has set up single sign-on. The first time you use Google, Facebook or your organization’s sign-in, you choose whether to create a new Jury account or to sign in to an existing one and link it. Matching email addresses do not automatically merge accounts. An organization that requires its own sign-in lends its seat and shared matters only to sessions started that way.

4. AI and other providers

Your question, relevant conversation context, and excerpts returned by search may be sent to OpenAI to generate an answer or conversation title. A search query may be sent to Google to create a semantic-search embedding. This processing may occur outside Mongolia.

The names and extracted text of documents, and the images selected for a question, are sent with that question to the configured OpenAI or Google AI service. An uploaded file is stored in your file library, and our server extracts document text or validates the image. When an image (JPEG, PNG or WebP) is uploaded to the file library, our server removes its metadata before storing it: EXIF data such as the camera, the time taken and the GPS location, XMP and IPTC data, comments and embedded thumbnails. Only the colour profile and the orientation that shows the picture the right way up are kept. The image is not otherwise changed; it is stored, downloaded and, when selected for a question, sent to the AI service in that form. Images uploaded to the file library before 13 October 2026 keep the metadata they were uploaded with until you delete them. Images sent through the older chat attachment upload, which the web chat no longer uses, are re-encoded without EXIF and other metadata and may be resized. The selected attachments are used for AI processing when you send the question.

To recognise text in scanned PDFs and images (OCR), the file is sent to the configured OpenAI or Google AI service, which is asked only to transcribe it. The recognised text is stored and marked as OCR text, which may contain errors. Public web research may send a search query and the address of a page to be read to Serper (serper.dev).

Google or Facebook sign-in opens that provider’s login page. The provider receives your request to sign in to Jury and technical information such as your IP address and browser details. Jury exchanges a verification code with the provider and checks the user identifier. This sign-in flow does not send your Jury conversations, files or billing records to the provider. Google’s and Meta’s own privacy terms also apply.

Organization sign-in opens the login page of the identity provider your organization chose (for example Google Workspace or Microsoft Entra). The provider receives your request to sign in to Jury and technical information such as your IP address. Jury exchanges a verification code with it and checks your identifier and email. Your Jury conversations, files and billing records are not sent to it. Your organization chooses and manages that provider, so its terms and your organization’s policies also apply.

When you send your case to an organization, the summary you agreed to, the information you selected, the conversation transcript if you chose it, and your contact details go to that organization or, where the channel’s organization works with a partner law firm that Jury has checked, to that partner. The receiving organization becomes the controller of the personal data in that copy and keeps it under its own responsibility; Jury stores and processes it on its behalf. Once it accepts, a lawyer there may copy the case into a research workspace in Jury and have AI research it. What you write in the conversation after sending does not reach the organization.

The organization that runs a channel or pays for it does not see your questions, answers or conversations; it sees only rounded totals. If you agree to topic statistics, it sees totals for closed months only, and no topic with fewer than 10 people; the text of your questions and your name do not go, and you can withdraw at any time. On a channel that admits people from a list, Jury does not keep the phone numbers and email addresses the organization uploads, only keyed digests of them, and the organization is the controller of that list. On a channel you join with a work account (single sign-on), the organization can see in its own sign-in system when you signed in.

verify.mn may be involved in phone verification, Bonum and QPay in payment, Postmark in email delivery, and Cloudflare and infrastructure providers in operating the site and network. Their processing is governed by applicable provider terms, settings, and law. We follow the principle of sending only the information needed for the relevant function.

We use Sentry to diagnose software errors, sending the error type, code location, application release, and environment. Before transmission, error reports exclude chat and document contents, request bodies, authentication information, and user identifiers. Session replay is disabled. Sentry receives network connection metadata, including the IP address, when a report is transmitted.

The chat app at chat.jury.mn loads the Meta Pixel, a script from Meta Platforms, on every page except those listed below, for every visitor, without asking first. Each time a page loads, your browser sends Meta a page-view report with your IP address, browser and device details, the address of the page (which may contain the internal identifier of a conversation or workspace, but never its content), the address of the previous page, the cookie the pixel sets in your browser and, if you are signed in to Facebook or Instagram in the same browser, Meta’s own cookies. Questions, answers, conversations, files, names, phone numbers and email addresses are not sent. The pixel does not load on administrator pages, an organization’s channel pages (/c/…), your cases (/cases), the organization console (/organization) or jury.mn. Meta may use this information under its own terms to measure ad results and to deliver and personalise ads, and may link it to your Facebook or Instagram account; the processing may occur outside Mongolia. The pixel does not load in a browser where you declined the ad-measurement question or turned it off in Settings; you can also limit it in Meta’s ad settings and with your browser’s tracking protection.

In addition, if you arrive from a Facebook or Instagram ad and accept ad measurement, our server reports these events to Meta Platforms through its Conversions API: the visit, a guest’s first question, a new account, and a confirmed payment with its amount. Each report includes your IP address, browser details (user agent) and Meta’s ad click identifier (fbclid). Questions, answers, conversations, files, names, phone numbers and email addresses are never sent to Meta. Meta may use this information under its own terms to measure ad results and improve ad delivery, and the processing may occur outside Mongolia.

5. Analytics and browser storage

The Google tag loads on every page with analytics storage denied by default. Before you choose, or after you decline, it does not create analytics cookies; Google may receive limited cookieless measurement signals containing information such as the page URL, IP address, browser or device information, and consent state. Accept enables analytics cookies. Advertising storage, ad-user data, and ad personalization always remain disabled. Your choice is stored in browser localStorage.

This landing site serves the Lora and Manrope fonts from jury.mn rather than Google. The chat app stores authentication tokens in localStorage; log out when using a public or shared device.

An essential, secure cookie inaccessible to JavaScript binds a social or organization sign-in request to the initiating browser for up to ten minutes. Provider access tokens are not retained in browser storage or the Jury database. Jury’s own session tokens continue to use localStorage.

When the chat app is opened from a link with campaign tags such as utm_source or utm_campaign, it keeps those tags and a random browser identifier in localStorage for up to 90 days and uses them to count visits, first guest questions and new accounts per campaign. This measurement stays on Jury’s own servers; it does not store IP addresses, browser details or Facebook and Google click identifiers, and it is not shared with third parties. Meta ad measurement is described separately above and below.

When you arrive from a Facebook or Instagram ad, the chat app asks once whether Meta ad measurement is allowed. If you accept, Meta’s ad click identifier (fbclid) is kept in this browser’s localStorage for up to 90 days and used to report the visit, a guest’s first question, a new account and a confirmed payment to Meta. If you decline, our server reports nothing, the Meta Pixel stops loading in this browser, and only your choice is kept; the page view sent when the page loaded cannot be recalled. Jury’s server keeps only a record that an event was reported; it does not store the IP address, browser details or click identifier. You can withdraw consent at any time from the chat app’s home page or Settings; that also turns the Meta Pixel off, and it does not affect reports already sent. The Meta Pixel keeps a random browser identifier in a cookie named _fbp for up to 90 days.

6. Retention and security

Files are stored privately in Cloudflare R2. Our database retains file metadata and extracted text. A file you upload is kept in your file library until you delete it: deleting a conversation or research workspace only removes its link to the file. A file moved to the trash can be restored for 30 days and is then deleted permanently; deleting it from the trash deletes it at once. Attachments uploaded outside the file library, through the older chat attachment upload, expire after 24 hours if never used in a question, and otherwise remain until their conversation is deleted. Deletion revokes access immediately and queues removal of the R2 object. Deselecting a file from a question does not delete it from the server.

Quoted excerpts and filenames may remain in conversation history. Removing an attachment does not immediately remove quotations in answers, backup copies, or copies retained by AI providers.

Signed-in users can start an incognito chat («Нууц яриа»). Its questions and answers are not saved to conversation history, it gets no title, and it does not appear in search or in the administrators’ conversation view. To prepare answers and give the next question its context, the server keeps the chat’s questions, answers, research steps, quoted excerpts and model context only while the chat is in use: they are erased when you end the chat, move to another conversation, sign out or close the page, and at the latest 24 hours after the last question or 7 days after the chat began. An incognito chat cannot upload files, create documents, save or report answers, or link to a research workspace; files chosen from your library are only read. Usage-meter records (tokens, cost, model) are kept without content. Erasure does not reach backup copies made earlier or AI providers’ own retention.

The organization keeps the copy of a case you sent for the period it sets, by default 3 years after the case closes (it can choose from 30 to 3,650 days). A request nobody answers expires after 7 days by default. When you withdraw a request before it is accepted, or it is declined or expires, the summary, the copy and your contact details are deleted and only a record that the request existed remains; they may stay in backups until those expire. An accepted case with no activity for 365 days closes. A list an organization uploads is deleted when it is replaced, when the channel stops using a list, or when the channel closes.

Retention depends on the information category, service needs, security, accounting, and legal obligations. Conversation, account, payment, audit, security, and accounting records may be retained for as long as needed for those purposes.

Logging out does not erase the account or server-side history. Using a history control, closing an account, or making a request does not mean every payment, accounting, audit, security, backup, or provider copy is automatically erased. No online service can guarantee absolute security.

Temporary social sign-in state expires after ten minutes and is cleaned up every minute while the sign-in service is running. Linked provider identifiers remain until a request to remove the link or account has been processed.

Organizations can request Jury’s data processing agreement at info@jury.mn.

7. Your rights and requests

Subject to applicable law, you may ask whether your data is processed and who received it, request a copy or correction, seek deletion or restriction where legally available, withdraw consent, or request review of an automated-processing result. Email info@jury.mn. We will verify your identity and assess any legal retention duty before acting.

You can delete your Jury account yourself with «Бүртгэл устгах» under Settings in the chat app and the mobile app. Deletion removes your conversations, saved sources, research workspaces and notes, file library, attachments and sign-in methods, and ends every session at once. Payment, invoice and usage-meter records and organization audit entries are kept for accounting and tax duties, linked only to an internal number without your phone number. Remaining plan time and Гэрэгэ packs end and any account balance can no longer be used; refunds are handled on request under the Terms of Service. An organization owner with other members must remove them first.

Deleting your account does not delete the copy of a case an organization has accepted: that organization is responsible for it. Deletion withdraws your pending requests and deletes their copies, and the organization is told on each accepted case that you deleted your account. Send requests to delete or correct that copy to the organization directly.

To request removal of a Google or Facebook link, sign-in data, or to have our staff delete your Jury account, email info@jury.mn and identify the provider and the phone number or email address on your Jury account. Do not send passwords, SMS codes or access tokens. Removing a link does not itself erase the Jury account, conversations or billing records. Revoking Jury’s access in Google or Facebook does not automatically delete information held by Jury; send a separate deletion request.

8. Contact and changes

For privacy, consent, analytics, or data requests, contact Смарт интел ХХК at info@jury.mn, 88110921, or 11th khoroo, Sukhbaatar District, Ulaanbaatar 14190, Mongolia. If the purposes, data categories, recipients, or retention terms materially change, this page will be updated and new consent will be sought where required by law.

On 12 November 2026 this policy added organization channels and sending a case to an organization. The app asks for your consent separately each time you send a case.

CONTACT

Смарт интел ХХК
Registration number: 6986897
Address: 11th khoroo, Sukhbaatar District, Ulaanbaatar 14190, Mongolia
Phone: 88110921
Email: info@jury.mn

OFFICIAL SOURCES
Law on Personal Data Protection↗Sensitive-data technical security regulation↗Automated-processing assessment regulation↗How Google uses information from sites using its services↗
Check information against its sources.
Product
About Jury GraphHow it worksDiscuss a pilotFor teamsJury AI legal assistantLaw library (Mongolian)
Contact
info@jury.mn88110921Facebook ↗Instagram ↗
Information
Terms of servicePrivacy policySource verification guideFAQ
© 2026 Смарт интел ХХК · Jury Graph · jury.mn
Information, not legal advice.