What actually happens when you click “Get feedback”
- Your file is uploaded over an encrypted (TLS) connection to our cloud storage.
- We extract the text and send it, along with your rubric, to a leading AI provider whose API runs the rubric grading and returns structured feedback. We choose providers whose enterprise API terms prohibit training on customer data.
- We turn that feedback into your report, save the report (so you can re-read it), and delete the underlying draft and rubric from our active storage within 3 hours.
- The AI provider does not store your input or output to train models. Their enterprise API tier prohibits training entirely. (The training-on-data behaviour you may have heard about is from consumer chat products - not the enterprise API tier we use.)
For institutional buyers
If you're evaluating Rubrica on behalf of a school, university, or organisation, we'll provide the specific list of AI sub-processors we use, copies of their Data Processing Agreements, and our internal data flow diagram under NDA. Email us via the contact page and we'll respond within one working day.
Our full Privacy Policy covers the legal detail on data retention and third-party processing.
What we don't do
- We do not keep a training corpus of student work. We have no internal model that learns from your essays.
- We do not sell, license, or share your draft with any third party other than the AI provider that grades it.
- We do not report your submission to your school, your institution, or any plagiarism database.
- We do notuse your work in marketing, in product demos, or as “examples” without explicit written permission.
What you should still be careful about
Even though your work is treated with care, a few small habits keep your data even safer:
- Strip personal informationyou don't need graded - your full address, phone number, your professor's personal email, anything sensitive. The grader doesn't need it.
- Use a cover sheet your rubric needs, not your full identity.Most rubrics don't require a real name or student ID to grade against.
- If you're unsure whether something is OK to upload, ask us first at contact.
How we're strengthening this further
We are in the process of putting Zero Data Retention (ZDR) agreementsin place with our AI providers. ZDR drops the provider's temporary 30-day abuse-review window down to zero - meaning your draft and the feedback exist on the AI provider's side only for the time it takes to compute your response, and are then immediately discarded. We will update this page when it is in effect.
We're also evaluating regional inference (running grading entirely inside a single cloud region for data-residency reasons) and a self-hosted enterprise tier for institutions that require it.
Questions?
We'd rather you ask than worry. Email us at contact and we'll answer within one working day.
TL;DR
Your draft is graded by leading AI providers under enterprise contracts that prohibit training on your work. We delete your file from our servers within three hours. We never sell, share, or report what you upload.