ChatGPT Integration Privacy Policy

Last updated: October 10, 2026

This policy applies specifically to the Be Underwater integration with ChatGPT. The data controller is EJ Digital LLC, 30 N Gould St Ste 5328, Sheridan, WY 82801, United States. Contact us at info@beunderwater.com. Website accounts and bookings are covered separately by our website privacy policy.

1. Scope and information received

The integration searches public diving activities, retrieves their details and checks calendar availability. It does not require a Be Underwater account, connect to your booking history, create reservations, process payments or provide medical assessments.

  • Searches: destination, optional activity category, maximum budget in USD and preferred language, and pagination settings (limit and offset).
  • Activity details: the selected activity identifier.
  • Availability: activity identifier, requested date and number of participants.
  • Technical data: the requesting connection's IP address and request timestamps for abuse prevention; tool name, success or failure and execution time for diagnostics. The connecting IP may belong to OpenAI infrastructure rather than your device.

ChatGPT sends tool arguments derived from your request. We do not request your full conversation, name, email, payment details, identity documents, medical records or certification documents through these tools. Text search fields can contain information you provide, so do not include sensitive personal information in them. The integration does not use website advertising or analytics cookies or link tool calls to a website account.

2. Purposes and information returned

We use the inputs to find activities matching your criteria, retrieve public details and calculate remaining places. Technical data supports request limiting, reliability and troubleshooting. The tool implementation does not use these requests for advertising, profiling or model training.

Responses sent to ChatGPT contain public activity identifiers, links, titles and descriptions; city and country; the dive center's published name and certification agency; prices, currencies and stored USD estimates; duration, number of dives, languages, inclusions and exclusions; published age, swimming, certification and experience requirements; and refundability and center-approval information. Availability responses also include the requested date and participant count, remaining places, availability status and the time of the check. We use existing reservation records internally to calculate remaining capacity, but do not return customers' identities or individual reservation records to ChatGPT.

3. Recipients

OpenAI receives our tool responses to provide answers in ChatGPT. Hosting and infrastructure providers process requests and operational data to operate the service; authorized personnel may access operational records for support and security. Tool calls do not send your enquiry to dive centers or payment providers. We may disclose data when required by law. We do not sell personal data.

OpenAI handles your conversation, tool arguments and results under its own privacy policy and your account settings. If you follow an activity link and use our website, the separate website privacy policy applies to its account, cookie and booking practices. Checking availability through ChatGPT does not make a booking.

4. Storage and retention

The tool handler processes arguments and results in memory for the request and does not save a search history or tool payloads to our application database. There is no persistent MCP session. Its diagnostic entries contain the tool name, success or failure and execution time rather than the arguments or result payloads. Request-limiting IP addresses and timestamps are held separately in server memory. Database errors and infrastructure request logs are separate from these tool diagnostics.

  • Tool arguments and results: processed for the duration of the request, without application-database persistence or a saved conversation history.
  • Request-limiting records: IP addresses and request timestamps remain in memory until the server process ends or an inactive entry is removed by memory cleanup. The limiter evaluates a 60-second request window; that window is not a deletion deadline. There is currently no guaranteed time-based expiry for these memory entries.
  • Operational diagnostics: application, database and proxy container logs rotate at 10 MB per file with up to three files retained per container. The oldest file is removed as new logs exceed that limit, or logs are removed with their container. Retention therefore depends on log volume and container lifetime, rather than a fixed number of days; low-volume records may remain until the container is removed.

The production proxy does not currently record every HTTP request in access logs, and application log forwarding to an external log-drain service is disabled. Database error diagnostics can include failed query details; successful database queries are not routinely logged. These diagnostics are used for operations and security, not marketing. Public catalog records and existing website reservations are maintained separately; checking them does not create a new booking record. OpenAI retains conversation and tool data under its own policies and your account settings, independently of these server retention rules.

5. Your controls

You can stop using or disconnect the integration in ChatGPT to stop further tool calls. Contact info@beunderwater.com to request access, correction or deletion of personal data held by Be Underwater. Because tool calls are not linked to a Be Underwater account, we may need limited information to locate relevant records; do not send passwords, payment details or medical records. Use ChatGPT's data controls for conversations and data held by OpenAI. Disconnecting alone does not delete existing records.