SPITSBERGEN LIBRARY · STUDY COMPANION
Privacy &
your study.
How the ChatGPT companion’s library connection handles information.
What reaches the library
When ChatGPT uses the Spitsbergen Library plugin connection, it sends a short search query, optional book or collection filters, or a passage identifier to spitsbergenlibrary.com. That service retrieves evidence from the public library service hosted at totbible.org and returns matching text, authors, titles, editions, and source references to ChatGPT. The earlier private custom GPT connects directly to that same TOT retrieval service.
Search terms can reflect religious interests. The connection is instructed to send only what is needed for a search, without names, confidential pastoral details, full private drafts, or the rest of your conversation. Do not include information that identifies you or another person when a general subject search will do. Do not submit passwords, payment data, government identifiers, or private health or counseling records.
The connection does not request your email address, a library account, or access to your private Drive. It cannot edit the library or read master documents that have not entered the public index. A forthcoming edition already in the public library is labeled as forthcoming.
How information is used and who receives it
The library uses the requested terms and identifiers to find relevant passages. Its hosting providers process the network requests needed to operate the .com connection and the TOT retrieval service. ChatGPT receives the results and generates the response in your conversation. This retrieval connection does not call the website’s paid AI-answer generator.
If you separately contact support, the people handling your request receive the information you choose to send. Do not include confidential details that are unnecessary to resolve the problem.
Storage and retention
The library connection’s application code does not store search-query text or ChatGPT conversation transcripts. Search text is sent in a request body, not a URL.
For capacity management, the retrieval service keeps one aggregate minute identifier and request count. These contain no query text, user identifier, or conversation identifier. The first accepted request in a later minute replaces the previous count. The last aggregate record remains while the service is idle; there is no scheduled expiration for it.
The .com connection records application failure labels for troubleshooting: a tool name and service status, or a generic failure label. These application messages do not contain search text or returned passages. They currently have no automatic expiration configured by the application.
Hosting providers may separately retain access and error logs containing network addresses, request times, paths, user agents, and service status for operation and security. For archived access logs on the spitsbergenlibrary.com hosting account, cPanel is configured for 30-day retention.
That 30-day setting does not cover the TOT retrieval host’s logs, PHP error logs, provider-internal security records, or backups. We have not configured or verified a fixed deletion period for those records; their retention follows the applicable hosting arrangements and provider policies. Disconnecting the plugin does not remove existing hosting records.
Support correspondence is kept until the issue is resolved and as needed for related follow-ups. There is no automatic deletion schedule for support messages. Use the support contact below to ask about information associated with your request or to request applicable access or deletion help.
Your choices
Use general subject terms and leave out personal details. You can stop using or disconnect the plugin in ChatGPT to stop new requests through that connection. Disconnecting it does not delete your ChatGPT history or existing hosting or support records.
Visit Study Companion support to ask about information held by the library service or to request applicable access or deletion help. Do not include more identifying information than is needed to locate your request.
On the website, the answer-feedback control copies a report to your clipboard. It does not automatically send that report. You control whether and where you share it.
Your ChatGPT account and conversations
OpenAI independently handles your conversation, uploaded files, and account information under its policies and your account settings. This policy describes the library retrieval connection, not OpenAI’s retention or account controls. The separate Ask the Library website generates answers through its own service; this policy does not describe that website’s full operation.
Questions
Email Cade@abidingplace.org or visit Study Companion support for questions about this connection.