Ultorn stores the data you give it directly: user accounts (email, hashed password, role), client records, scan scopes (domains/IPs/URLs you add), scan results and findings (including evidence such as HTTP headers, banners, and page content captured during a scan), and an audit log of actions taken in the platform.
When you use the AI Guard feature, your prompt is sanitized to strip client names, IPs, credentials, and other sensitive patterns before it is sent to an external model (or kept local if it is still sensitive, or if you request local-only processing). We log the provider, model, and a one-way hash of the prompt and response, not the raw text.
In the default self-hosted configuration, all data is stored in a local SQLite database under your own infrastructure. Nothing is sent to Utachi Industries or any third party except the AI provider you configure, and only for the sanitized portion of a query you explicitly submit.
Access is role-gated: clients see only their own clients/scans/findings, analysts and admins see all records, and only admins can view the audit log, manage users, or manage API keys. API keys authenticate as the admin account that minted them.
Records persist until you delete them at the infrastructure level (there is no automatic expiry in this MVP). Because the audit log is immutable by design, actions remain logged even if the underlying record is later removed.
This is a plain-language description of what the current Ultorn codebase actually does with data. It is not a substitute for a lawyer-reviewed privacy policy for your jurisdiction or client contracts, and should be reviewed before commercial use.