DILAbot
+

Privacy Policy

Last updated: 2026-11-14

DILAbot (“the Service”) is provided by Dharma Drum Institute of Liberal Arts (“DILA”, “we”). In line with Taiwan's Personal Data Protection Act, this page explains what data the Service collects, how it is used and kept, and your rights. The rules for using the Service are in the Terms of Service. If this English version differs from the Chinese version, the Chinese version prevails.

1. Purposes

To provide the Buddhist question-answering service, manage accounts and daily free credits, compile usage and cost statistics, keep the system secure, enforce the Terms of Service and prevent abuse, and improve answer quality.

2. What we collect

  • Account data: when you sign in with Google, the account identifier, email, name and profile picture URL that Google provides. We never receive your Google password.
  • Conversations: your questions, the model's answers, the retrieval steps (including search terms), and any titles you give your conversations.
  • Usage records: for each question, the time, reasoning effort, names of the tools used, token usage and cost.
  • Problem reports: the category and comment you submit with “Report a problem”, together with a copy of that question and answer.
  • API keys you register: OpenAI, OpenRouter or Anthropic API keys, stored encrypted. Neither administrators nor maintainers can see them on screen.
  • Settings: interface language, appearance, answer model, selected tools and similar preferences.
  • Enforcement records: when an account is suspended or terminated for violating the Terms of Service, the time, the reason and the conversations concerned.
  • System logs: the server records request times, URL paths and IP addresses for debugging and security. The text of your questions is not written to these logs.

3. Who processes your data

To produce answers, your questions and conversation are sent to the following services:

  • OpenRouter (United States) and the model providers it routes to: answer models such as DeepSeek, and ChatGPT or Claude if you registered an OpenRouter API key but no OpenAI or Anthropic key. We only allow providers that neither retain requests nor train on them. DeepSeek questions are not handled by providers based in China. Before each answer, the text of that question (without earlier messages or account data) is also sent to a classification model on OpenRouter to decide whether a brief or a detailed answer fits.
  • OpenAI (United States): the ChatGPT answer model. Data is handled under OpenAI's API data policy and is not used for model training by default.
  • Anthropic (United States): the Claude answer model, called directly only when you registered your own Anthropic API key and picked Claude (with only an OpenRouter API key, Claude goes through OpenRouter). Data is handled under Anthropic's API data policy and is not used for model training by default.
  • DeepInfra (United States): converts search text into vectors for semantic search.
  • NTU Digital Library of Buddhist Studies (Taiwan): when the assistant searches the research bibliography, the search text (without your account details) is sent to National Taiwan University's bibliographic search service.
  • Google: sign-in; pages also load fonts from Google Fonts.

If you register your own API key, your questions are sent with your key to your own OpenAI, OpenRouter or Anthropic account, which pays for them. The brief-or-detailed check still goes through the Service's OpenRouter account, at the Service's expense.

The Service's maintainers (DILA staff) can view conversations, retrieval steps and problem reports to fix bugs, improve answers and investigate violations of the Terms of Service. We do not sell your data and do not use your conversations to train AI models.

Share links: when you share a conversation, anyone with the link can read it up to the point you shared, without signing in. Shared pages ask search engines not to index them. You can stop sharing at any time and the link stops working.

4. Where and how long data is kept

  • Data is stored on DILA's servers in Taiwan. Data sent to the third-party services above is processed in the countries where those services operate.
  • Your account and conversations are kept until you delete them. Deleting a conversation removes its questions, answers, retrieval steps and share link; the usage records for its questions (without any text) are kept for statistics.
  • Deleting your account removes your account data, all conversations, problem reports and API keys; usage records are kept but no longer linked to anyone. We also keep a one-way hash of your Google account identifier, so that the account cannot be re-created the same day to claim another day's free credits; it is deleted the next day.
  • If your account is suspended for violating the Terms of Service, the account and its conversations are kept until the matter is resolved, for review and appeal. If it is terminated, data is deleted as when you delete your account, except that we keep a one-way hash of your Google account identifier and the reason for termination, so that the same Google account cannot register again; this record is kept until DILA lifts it.
  • The temporary buffer used to stream answers to your screen keeps them for at most one day before clearing them automatically.
  • Deleted data may remain in system backups until those backups are rotated out.

5. Cookies

The Service only uses cookies needed for it to work. There are no advertising or tracking cookies:

  • Sign-in session.
  • dila_bot_locale: interface language.
  • dila_bot_theme: light or dark appearance.
  • dila_bot_sidebar: whether the sidebar is collapsed.
  • dila_bot_disabled_tools: retrieval tools you have turned off.

6. Your rights

You may ask to see your personal data, receive a copy, have it supplemented or corrected, have its collection, processing or use stopped, or have it deleted. You can delete conversations yourself at any time, and delete your account from the Settings page. For anything else, please email us. If your account is suspended, you can still ask by email to have it deleted; the record used to refuse re-registration described above is kept after deletion.

You may choose not to provide data, but asking questions requires signing in with a Google account. Without signing in you can still browse the home page and the tool descriptions.

7. AI answers

Answers are generated by AI models and may contain mistakes. Rely on the cited primary sources. Please do not include sensitive personal data about yourself or others in your questions.

8. Changes

When this policy changes, we update this page and its “Last updated” date.

9. Contact

Dharma Drum Institute of Liberal Arts
Email: ai@dila.edu.tw