Government and Public Sector / AI Transparency Statement

AI Transparency Statement

First Touch Studio publishes this statement to the same standard Australian government agencies publish their own. It sets out where we use artificial intelligence, where we do not, how we handle your information, and how you can hold us to it.

Version
1.0
Effective
01/7/2026
Next review
1/12/2026
Accountable officer
Marcin, Founder
01

Purpose and scope

This statement applies to all work First Touch Studio delivers, in every jurisdiction, and to every tool we use in delivering it. It exists because you are accountable for what your suppliers do with your information. Governments across Australia now require assurance that third party suppliers using AI meet their privacy and information security standards, and that obligation does not stop at the agency boundary. It reaches us.

We publish it rather than waiting to be asked. If you are evaluating us, this document should answer your AI questions before you have to raise them. Where your jurisdiction applies a specific framework or risk assessment, tell us which, and we will complete it. If this statement does not answer something your process requires, ask, and we will answer in writing.

Who this covers

First Touch Studio (ABN 27 059 706 024) and any subcontractor engaged on your work. Subcontractors are bound to this statement by written agreement before they touch your material.

02

Our position

We use AI. We are not going to pretend otherwise, and a supplier who tells you they do not is either behind or not being straight with you.

What we do not do is let it near your judgement calls, your community's data, or anything that reaches the public without a person having written and checked it. AI makes production faster. It does not make decisions, form strategy, or carry accountability. Those remain human, and they remain ours.

The distinction we hold to is simple. AI can help us do the work. It cannot do the work.

03

Where we use AI

We use general purpose generative AI tools, on business tier agreements, in the following areas. This list is exhaustive. If a use is not named here, we are not doing it.

Drafting and editing assistance
First drafts, restructuring, tightening and proofreading of material we have already conceived and briefed. A human writes the brief, sets the argument, and rewrites the output.
Human-authored, AI-assisted
Summarisation of material you have provided
Condensing long documents you have given us, where that material has been assessed as suitable and, where required, de-identified first.
Approved material only
Production variants and formatting
Adapting approved copy across formats and channels, resizing and reformatting assets, generating alt text drafts for accessibility review.
Post-approval only
Internal research and orientation
Getting up to speed on a sector or topic using publicly available information. Nothing produced this way is presented to you as evidence without independent verification against a primary source.
Verified before use
Internal productivity
Our own scheduling, note-taking and administration. Your material is not involved.
No client data

Every one of these is a production aid. In each case a person has authored the thinking, and a person reviews the output before it goes anywhere.

04

Where we do not use AI

These are not preferences. They are commitments, and you are entitled to hold us to them.

We do not use AI to make, inform, or recommend any decision that affects a member of the public.
We do not use AI to generate community consultation or engagement material without human authorship and review.
We do not use AI to produce research findings, statistics, quotations or community sentiment. If we present something as evidence, a person has sourced and verified it against a primary source.
We do not use AI to write strategy, positioning, or recommendations. That is the judgement you are paying for.
We do not use AI image generation to depict real people, real places, or events that did not happen.
We do not use AI to analyse or profile individuals, communities, or submissions in a way that affects how they are treated.
We do not deploy AI tools that interact directly with the public on your behalf without your explicit written approval and a joint risk assessment.
We do not charge you for AI-generated volume and describe it as strategic work.
05

Information handling

This is the section that matters most, so it is the most specific.

What never goes into an AI tool

Personal information or any material from which an individual could be identified.
Official information, as defined by the agency providing it.
Commercial-in-confidence material, whether yours or a third party's.
Anything marked confidential, protected, or subject to a security classification.
Unpublished consultation submissions, community feedback, or complaint data.
Anything you have told us not to.

Before any AI-assisted step

Material is assessed against the list above. Where it is suitable but contains identifying or sensitive detail that is not needed for the task, that detail is removed first. The tool sees only what it needs to see, and often that means it sees nothing at all.

Tool terms

We use business tier agreements only. Consumer and personal plans are not used for any client work. Our tools operate under terms that provide, at minimum, that our inputs are not used to train the vendor's models, that data retention is defined and limited, and that confidentiality is contractual rather than a matter of vendor goodwill. We will tell you which tools are in use on your engagement, and provide their current terms, on request.

Data location

We do not transfer client, community or citizen information offshore without your written approval. Where a tool processes data outside Australia, that is disclosed in the register and we will not put your material through it without asking you first.

06

Human oversight and accountability

One named person is accountable for AI use on your engagement, and you will know who they are before work starts. For most engagements that person is the Principal.

Every deliverable is reviewed by a person before it reaches you. Every public-facing deliverable is reviewed again before it reaches the public. There is no path by which an AI output travels from a tool to your community without a human having read it, checked it, and taken responsibility for it.

Where AI has assisted in producing something we deliver to you, we tell you. In writing, in the deliverable, without being asked. You should never discover it later.

07

Framework alignment

This statement and our underlying practice are aligned to the following. We did not invent our own principles, because you should not have to evaluate a supplier's homemade ethics. Where your jurisdiction applies its own AI governance policy or risk assessment, we work to that as well.

Australia's AI Ethics Principles
Department of Industry, Science and Resources. Explicitly intended for adoption by private businesses as well as government.
State and territory AI governance policies
Including the Queensland Government AI Governance Policy (QGEA). Where your jurisdiction has its own, we work to it. Tell us which applies and we will confirm alignment in writing.
ISO 38507
Governance implications of the use of AI by organisations. The international standard that underpins several Australian jurisdictions' AI governance requirements.
DTA Standard for AI transparency statements
The Commonwealth standard this document is modelled on.
AI risk assessment
We will complete your agency's AI risk assessment on request and at no cost, whichever framework it sits under. We are familiar with FAIRA and with the Commonwealth assurance framework.
Privacy obligations
The Privacy Act 1988 (Cth) and the applicable state or territory privacy legislation for the material you provide.
08

Assurance and audit

Your agency is obliged to assure itself that third party suppliers using AI meet your privacy and information security standards, and to audit those suppliers. That is true in every jurisdiction we work in. We would rather you did audit us than did not.

What we will provide, on request, at no cost

A completed AI risk assessment against whichever framework your jurisdiction applies. A current list of the tools in use on your engagement, with vendor terms attached. A record of where AI assisted on your engagement. Evidence of our tool plan tiers and contractual terms. Access for you or your auditor to inspect any of the above.

If an audit finds a gap, we will tell you what it is, fix it, and tell you when it is fixed. We will not argue the point.

09

Incidents and breaches

If information is entered into a tool in error, or a vendor suffers a breach affecting material of yours, we will tell you. Not eventually. Immediately, and in writing.

We will tell you what was involved, when, what the exposure is, and what we are doing about it. Where the incident is a notifiable data breach, we will support your agency in meeting its obligations under the applicable scheme and we will not contest the assessment.

We would rather report an incident that turns out to be nothing than fail to report one that turns out to be something.

10

Records

Where AI has assisted in producing material we deliver to you, we keep a record of it: what was produced, which tool assisted, what input was provided, and who reviewed the output. That record is retained for the life of the engagement and provided to you on request.

Material we deliver to a government agency may become a public record. We treat it that way from the outset.

11

Review and contact

This statement is reviewed at least every six months, and immediately whenever we adopt a new tool, a vendor materially changes its terms, or the governing frameworks change. Each version is dated and the previous version is available on request.

Accountable officer

Marcin, Founder, First Touch Studio
hello@firsttouchstudio.com.au · 02 6190 7977
Questions about this statement, our AI use, or a specific engagement are answered in writing.

If you are evaluating First Touch Studio and this statement does not answer a question your process requires, tell us what you need and we will provide it. We would rather do the work than have you guess.