Zero Data Retention Is Now Enabled on Our OpenAI API Workloads
Since October 8, 2026, OpenAI does not retain the data Ur AI sends through its API. It is one more step in our commitment to data sovereignty for enterprises.
Since October 8, 2026, Zero Data Retention (ZDR) has been enabled across Ur AI's OpenAI API organization. When our products send a customer's documents to OpenAI's models, OpenAI does not keep that content or the outputs after the request is processed.
What changes
By default, OpenAI's API may keep request data for up to 30 days for abuse monitoring. Under ZDR, that content is not retained, so nothing from your documents stays with the model provider after the request is processed.
ZDR is set at the level of our OpenAI organization, and every project inherits it, including new ones. It covers the work we do for customers on the OpenAI API:
- Nebula, our document intelligence engine, which turns PDFs, scans, tables and charts into structured data and answers questions on them.
- Specter, our M&A due diligence product, which analyzes confidential financial, legal and business documents and drafts diligence reports.
- Custom workflows built on Nebula and Specter, such as cross-document analysis and tailored data extraction.
As before, OpenAI does not use data sent through its API to train its models.
Why it matters
Our customers work with documents that must never leak: deal materials, financial statements, contracts and internal records. For financial institutions, M&A teams and regulated industries, the question is not only whether AI is accurate, but where the data goes and who keeps it. With ZDR in place, using a frontier model no longer means leaving a copy of your data behind.
I believe enterprises should be able to use the best models without giving up control of their data. ZDR closes one more gap, and it is part of a simple principle for us: your data stays where you decide.
Part of our sovereignty commitment
ZDR is one layer of a broader approach. Ur AI is ISO 27001:2022 certified and has completed a SOC 2 Type 2 examination and its GDPR compliance program. For workloads that cannot leave a customer's environment at all, Nebula Sovereign runs on open-weight models that we post-train ourselves and can be deployed in a private cloud or on-premises.
Frontier models when you want them, sovereign models when you need them, and in both cases your data under your control.
If your team needs AI that meets strict data requirements, we would like to hear from you. Get in touch at contact@ur-ai.net.
Frequently Asked Questions
What is Zero Data Retention (ZDR)?
ZDR is an arrangement with OpenAI under which the content sent to eligible API endpoints, and the outputs, are not stored by OpenAI after the request is processed, including for abuse monitoring.
Since when is ZDR enabled for Ur AI?
Since October 8, 2026, across Ur AI's OpenAI API organization.
Does it apply to all Ur AI products?
It applies to the Ur AI products and custom workflows that use the OpenAI API, including Nebula and Specter. Nebula Sovereign, which runs on Ur AI's own post-trained models, does not send data to OpenAI at all.
Do I need to do anything as a customer?
No. ZDR is applied on Ur AI's side and needs no change to how you use our products.
