Trust
AI and your data
The AI agent (part of Max; not available yet) helps you ask questions and draft queries and scripts. Here is what it does with your BUSY data, and the one case in which a data sample reaches an AI provider.
In normal use
- The agent builds queries and retrieves results from your BUSY data; it does not send your accounting data in full to a third-party AI provider.
- It drafts reports, reminders and scripts for you. A drafted script starts inactive; you preview, test and activate it.
- It never activates a script or posts a voucher by itself. You authorize every posting.
The debugging exception
If a query the agent built returns a wrong result, the agent can ask to use a sample of the affected data to correct it.
- 1
Step 1: You are asked
Nothing is sent unless you give explicit consent for that correction. Your consent decides whether a sample is sent; it does not change OpenAI’s own retention or training settings. - 2
Step 2: A small sample is sent
Only a small, relevant sample of the affected data is sent to the OpenAI API to correct the query. - 3
Step 3: The query is corrected
You review the corrected query and its result as usual.
AI Zeny’s temporary copy
- AI Zeny deletes its own temporary copy of the sample after the query is corrected.
- You can delete that copy yourself at any time.
- A nightly cleanup removes anything left.
These steps delete AI Zeny’s copy only. They do not delete anything held by OpenAI.
What OpenAI keeps
What OpenAI keeps or does with a sample sent this way is governed by OpenAI’s own terms for its API. On 17 October 2026, OpenAI’s page Data controls in the OpenAI platform said:
- Data sent to the OpenAI API is not used to train or improve OpenAI’s models unless the OpenAI API customer — the organization whose OpenAI account sends the requests — explicitly opts in to share it.
- Abuse-monitoring logs may contain prompts and responses. They are normally kept for up to 30 days, with longer retention possible for legal or safety reasons.
- Some API features keep application state; retention depends on the endpoint, feature and settings. See the current endpoint table on OpenAI’s page.
- Zero Data Retention and Modified Abuse Monitoring require eligibility and OpenAI’s prior approval and have endpoint, feature and safety limitations.
Read it in context
This describes OpenAI’s API platform in general, not the consumer ChatGPT product, whose terms differ. The choices it mentions — opting in to share data, retention controls — belong to the OpenAI account that sends the requests; they are not choices you make in AI Zeny. Which OpenAI account, API features and settings AI Zeny’s debugging exception uses has not been published on this website yet. OpenAI may change its terms; its page is the authority.