Trust and vendor file AI use statement

AI use statement

Last reviewed: [DATE]

Innorve Academy teaches credit-union staff how to use and oversee AI. This statement explains where Innorve itself uses AI, and where it does not.

Summary

Question Answer
Does the platform use an AI model to process customer data? No. No AI model processes customer data in this version of the platform.
Does the platform send data to an AI provider? No. Nothing in the platform sends data to an AI provider.
Is customer data used to train AI models? No.
Was AI used to create the course content? Yes. Course materials were drafted with AI assistance and reviewed by people.
Is there an AI coach or tutor? Not in this version. A coach may be added later under the rules below.

The platform (version 1)

The platform stores and displays training records. It computes rubric totals and cohort aggregates with ordinary database logic. It does not use any AI model to:

  • score or assess learners;
  • generate feedback;
  • analyze submissions, pilot records or evidence-binder entries; or
  • make any decision about a learner or a credit union.

Reviewer scores and feedback are written by people. Rubric totals are computed by the database from the scores reviewers enter.

Course content

  • Course materials were drafted with AI assistance.
  • People review the content. Practitioner review of the content is in progress.
  • Each source the course cites is labelled by type: binding rule, statute, state law, contract, supervisory guidance, exam procedure or benchmark. See the not-legal-advice page.
  • The content is tool-neutral. It teaches how to use and oversee AI tools in general, and does not require or promote any particular AI product. Credit unions practice with the tools they have approved for their own use.

Rules for any future AI coach

If Innorve adds an AI coach to the platform, it will follow these rules. They will be built in and tested before release, and customers will be told in advance. [CONFIRM notice period; the draft DPA also requires written agreement before any customer data is sent to an AI provider]

  1. Approved sources only. The coach may draw only on approved learning sources. [CONFIRM who approves sources and how the list is kept]
  2. Formative hints only. The coach gives hints and questions to help a learner think. It does not write the learner's work for them.
  3. No unseen answers. Answer keys and unseen assessments are kept out of the coach's context, so it cannot reveal them.
  4. No passes, no actions. The coach cannot award a pass, change a score, release feedback or take any action in the platform or in any credit-union system. Assessment stays with human reviewers.

Any coach would also need:

  • a subprocessor notice under the DPA, if it uses a new provider;
  • an update to this statement, the subprocessor list and the data inventory; and
  • no use of customer data for model training.

What we ask of credit unions

  • Do not enter member data into the platform, including into any future coach.
  • Use only AI tools your credit union has approved when you practice outside the platform.
  • Remember that completing a course does not grant authority to use AI at the credit union.