Everyday AI data handling
Review AI drafts of security answers
Use AI assistance only with answer text and supporting material approved for the intended recipient and service. Keep the source, scope and revision of each answer visible, and withhold confidential evidence unless its owner permits that disclosure. A polished response does not turn a policy statement into proof that a control operates effectively.
For Sales engineering and security assurance
Review what supports the statement before sharing the evidence.
Entirely fictional supplier and policy. No real control approval, audit or certification is represented.01
Start from the supplier’s approved answer source
This guide is for the team answering a customer security questionnaire. Identify the current approved answer library or responsible control owner before drafting. A procurement team comparing vendors has a different task and should use the separate questionnaire template.
Record the question, proposed answer, supporting source and its revision. If the answer depends on a scope limit or an exception, keep it with the statement. Do not ask an assistant to fill an evidence gap with a plausible assurance claim.
02
Separate answer text from confidential evidence
An approved high-level answer may be suitable for a customer while an internal report, configuration export or access-review record is not. Review the audience and purpose of each supporting artifact independently rather than treating permission to answer as permission to upload the entire evidence pack.
The synthetic pack includes a small fictional policy, an answer derived from it and a restricted-evidence marker. The example deliberately labels the answer as a policy statement only. It does not claim an audit, certification, operating-effectiveness result or any actual Aona practice.
| Example material | What it supports | Exercise sharing decision |
|---|---|---|
| Fictional policy clause | Owner reviews demo access requests | Use the scoped example answer |
| Approved answer record | A traceable summary of that clause | Keep its source and limitation |
| Restricted evidence marker | Stands for a separate internal artifact | Withhold pending authorised review |
03
Keep limitations when using drafting assistance
An AI draft should preserve the original meaning, evidence level and permitted audience. Check that it has not turned an intended policy into a measured result, removed a scope limitation or implied a broader report than the source supports.
For the exercise, compare the source clause with the prepared response. Both describe only the fictional access-request process. The response retains the source reference and limitation so a reviewer can assess the statement without receiving the restricted evidence marker.
04
Review the AI input as a separate disclosure
Before submitting a questionnaire or evidence file to an AI service, review customer names, personal information, internal configuration details and confidentiality terms. A document used to answer a customer may still need minimisation before it becomes AI input.
OAIC guidance addresses the privacy implications of personal information supplied to AI. It does not approve the supplier’s assurance claims or determine contractual confidentiality. Use the organisation’s own authorised process for the actual service, account and evidence audience.
Source context: OAIC: Use of commercially available AI products
05
Release the answer through its owner
Close the response with the control owner’s approval, source version and any conditions on sharing. Keep a record of which evidence was withheld or shared through a different approved channel. If the underlying policy or report changes, update the answer rather than reusing an old polished draft.
The download contains a populated fictional example and a separate blank review record. No customer questionnaire has been answered or sent. Technical input protection is another layer and does not certify the accuracy of the submitted answers.
Put it into practice
Supplier answer and evidence exercise
Compare a fictional source clause, its scoped answer and a separate restricted-evidence decision.
Entirely fictional supplier and policy. No real control approval, audit or certification is represented.
| Record | Populated example | Limit |
|---|---|---|
| Question | Who reviews demo portal access? | Fictional questionnaire |
| Answer | The demo owner reviews access requests | Policy statement only |
| Source | demo-policy.txt, access clause | Synthetic source |
| Restricted evidence | Separate internal marker | Withheld in the exercise |
Work through your review
Use the checks to organise the evidence you need. Your selections stay in this tab.
0 of 3 reviewed
Example files for this task
Keep the source material and the instructions together. You can also download the complete worksheet or matrix as CSV.
README.mdInspect
# Supplier questionnaire response exercise
SYNTHETIC SUPPLIER ONLY. Nothing in this pack describes Aona or another real company’s controls. No questionnaire or evidence is sent to an AI service or customer.
Read demo-policy.txt, compare answer-library.csv and prepared-answer.md, then review the separate restricted-evidence-canary.txt decision. Keep the policy-only limitation and source reference. Use review-record.md for the actual owner process without copying confidential evidence into this training pack.
## Guide and sources
Canonical guide: https://aona.ai/resources/guides/security-questionnaires-ai-input-controls/
Source review: 2026-09-21
- OAIC: Use of commercially available AI products: https://www.oaic.gov.au/privacy/privacy-guidance-for-organisations-and-government-agencies/guidance-on-privacy-and-the-use-of-commercially-available-ai-products
demo-policy.txtInspect
SYNTHETIC_SUPPLIER_U08 POLICY
Revision: DEMO_V1
Access clause: The fictional demo portal owner reviews account requests before creating access.
Scope: the fictional demo portal only.
Evidence level: policy statement. No audit or operating-effectiveness result is supplied.
answer-library.csvInspect
question_id,question,scoped_example_answer,source_ref,exercise_audience,limitation
DEMO_Q1,Who reviews demo portal access?,The fictional demo portal owner reviews account requests before access is created,demo-policy.txt access clause,SYNTHETIC_BUYER_A,Policy statement only
DEMO_Q2,Can we receive the internal evidence file?,That separate artifact requires its own authorised disclosure review,restricted-evidence-canary.txt,NOT APPROVED IN EXERCISE,No evidence disclosure implied
prepared-answer.mdInspect
# Prepared fictional response
Supplier: SYNTHETIC_SUPPLIER_U08
Question: Who reviews demo portal access?
Answer: The fictional demo portal owner reviews account requests before access is created.
Source: demo-policy.txt, access clause, DEMO_V1.
Scope: fictional demo portal only.
Evidence level: policy statement, not an audit or operating-effectiveness result.
Audience in this exercise: SYNTHETIC_BUYER_A.
Restricted evidence: not attached or authorised by this example.
restricted-evidence-canary.txtInspect
SYNTHETIC_U08_INTERNAL_EVIDENCE_MARKER
This stands for a separately controlled evidence artifact in the exercise.
It contains no actual report, customer detail or configuration.
review-record.mdInspect
# Actual response review
Question and intended audience: ____________________
Approved source/revision/owner: ____________________
Answer and retained limitations: ____________________
AI service/account permitted for this input: ____________________
Evidence separately approved or withheld: ____________________
Final owner decision: NOT YET REVIEWED
Before you proceed
Keep these distinctions clear
- Answering beyond the source
- A policy, certificate or report supports a particular statement and scope, not every assurance a questionnaire requests.
- Uploading all evidence to improve the draft
- Review each artifact’s confidentiality and audience independently from the high-level answer.
Apply it to employee AI use
Bring your actual data path.
Aona can help evaluate supported employee prompt/file paths used while preparing a response.
Aona does not validate the truth of a questionnaire answer or authorise disclosure of confidential assurance evidence.
Use the fictional response pack in a scoped input-policy discussion while the control owner remains responsible for answer accuracy.
Review your use caseFAQ
Questions for this decision
Can the approved answer library be used without its source references?
Does permission to answer mean we can upload the full internal report?
Do the example answers describe Aona’s controls?
Can AI decide whether the control is effective from a policy statement?
Evidence behind the guide
Sources and scope
Prepared by Aona. Sources checked 2026-09-21. The cited material supports the specific points below; it does not certify a product or your use case.
- OAIC: Use of commercially available AI products
Explains privacy review for personal information submitted to AI; does not establish supplier assurance claims.
regulator · checked 2026-09-21