Privacy
- Our Approach
- What the API Does Not Keep
- What the API Keeps
- Where Documents Are Processed
- What You Should Do
Our Approach
Identity documents hold very sensitive personal data. IDParse is built so that the API keeps as little as possible. A document passes through one request, you receive the answer, and the API keeps a billing record only.
This page describes what the API does today. It is a technical description, not a legal text. The privacy policy explains how we handle personal data. The legal terms are in our Privacy Policy.
What the API Does Not Keep
Document content is not saved in application records or logs. This is true for:
- the content of the document, which means the uploaded file,
- the file name,
- the extracted fields, such as names, dates and document numbers.
While a request is processed, temporary upload and processing files are used. The software is designed to remove them during cleanup.
Error monitoring (a tool that reports software faults to our developers) is switched off for scan requests. So a software fault during a scan is not reported to such a tool.
The notes in an answer are short technical remarks. They never contain document data.
What the API Keeps
For each scan the API keeps a billing record. We need it to count your usage, to bill correctly and to answer support questions. The record holds only:
| Item | Example |
|---|---|
| Account identifier and API key identifier | Internal numbers |
| Endpoint and HTTP method | v1/extract, POST |
| HTTP status | 200 |
| Whether the scan was charged | Yes or no |
| The quota the request was counted against | monthly or lifetime |
| A short outcome code | extracted, no_mrz_found |
| How many documents the engine returned, and how many of them passed the check digit checks | 2 and 1 |
| Whether the file was read completely, or a limit cut the reading short | Yes or no |
| File size | In bytes |
| Processing time | In milliseconds |
| Which engine server processed the scan, and which engines were asked with their result | Internal numbers and fixed words, such as served or busy |
| Timestamp | Date and time of the request |
The outcome code says how the scan ended. The counts say how many documents the engine returned. Neither says anything about the person or the document.
Where Documents Are Processed
We process documents on dedicated servers in Germany, hosted by Hetzner and operated by Wally Gallery OÜ. Results are returned to your application.
What You Should Do
- Send documents over HTTPS only, as the base URL of the API shows.
- Call the API from your own server. Never place an API key in browser code or in a mobile app.
- Store only the fields you really need, and only as long as you need them.
- When you contact support, give your account, the time of the request and the endpoint. Never send document images or extracted data.