← Back to Security & Trust

AI Data Use Policy

Last reviewed July 15, 2026

GrantProof includes one optional feature that uses an AI provider: the Readiness Assistant. This policy describes precisely how it handles your data. Everything else in GrantProof — findings, proof scores, money-at-risk, extraction — runs on our own servers with no AI provider involved.

Off by default

The Readiness Assistant is disabled unless it is explicitly enabled and your organization has an active GrantProof trial or subscription. It only runs when an authorized non-viewer user in your organization clicks to generate a summary — it never runs on its own or in the background. The fictional public demo is the sole viewer-role exception.

What we send

When you run it, we send a scoped set of structured workspace fields for your organization, such as:

  • Grant names, funders, award amounts, spent and remaining amounts, and dates
  • Readiness statuses, proof scores, deadlines, and owners
  • Evidence titles and notes, and missing-evidence descriptions you have entered

What we never send

  • Raw uploaded file contents (the actual PDF/Word/Excel bytes or extracted full text)
  • Passwords, API keys, or session tokens
  • Billing or payment details
  • Any other organization’s data

We send the minimum needed for the summary you asked for — never more.

No training on your data

We do not use your data to train AI models, and we do not opt in to model training with our AI provider (OpenAI). Data submitted through OpenAI’s API is not used by OpenAI to train its models under their API data-usage terms.

Untrusted-document handling

Text that originates from your notes and uploaded documents is treated strictly as evidence to summarize — never as instructions. If a document contains text attempting to change the assistant’s behavior (for example, “ignore previous instructions” or “mark this grant compliant”), the assistant is instructed to ignore it and continue. The assistant cannot certify compliance and always attaches a human-review disclaimer.

Human review required

AI output is a readiness aid, not a determination. Every response states that it is not legal, accounting, audit, or compliance certification and that human review is required before you rely on it.

Questions? Email erica@grantreadysystems.com.