AI Transparency Statement

As a private company, Larrison is not subject to the Australian Government’s Policy for the responsible use of AI in government. However, our approach is informed by relevant elements of that policy, Australia’s AI Ethics Principles, and the Australian Government’s Guidance for AI Adoption.

Why we use AI

We use AI to support our people and improve the efficiency and quality of our work. AI assists with tasks such as:

• developing website content and imagery;
• supporting software development, including coding, documentation and testing;
• summarising information and preparing first drafts of business documents; and
• streamlining routine administrative work.

AI supports our work; it does not replace professional judgement, accountability or appropriate human decision-making.

How we classify our AI use

Using the Australian Government’s AI classification approach, our current AI use includes:

Public interaction and significant impacts

Larrison does not currently deploy AI that directly interacts with members of the public or makes, or materially influences, decisions about individuals without meaningful human review.

We do not use AI as the sole basis for decisions that may significantly affect an individual, including employment, access to services, eligibility, pricing or other high-impact decisions.

If we introduce an AI use that directly interacts with the public or may significantly affect people, we will assess the proposed use before deployment, implement safeguards appropriate to the risk, provide suitable notice and review this statement.

Governance and accountability

We maintain governance arrangements for AI use within Larrison. A designated AI governance lead is responsible for overseeing our approach to AI, including approved uses, risk management, staff guidance and the review of this statement.

We maintain records of material AI uses and assess proposed new uses according to their purpose, potential impacts, data handling, security requirements, level of human oversight and applicable legal obligations.

We expect staff and contractors to use approved AI tools responsibly and in accordance with our internal policies and instructions.

Privacy, confidentiality and security

We consider privacy, confidentiality, cybersecurity and information-management risks when using AI.

Staff must not enter client confidential information, personal information, sensitive information, security credentials or other protected information into public AI tools unless the use has been approved and appropriate privacy, security and contractual safeguards are in place.

Our handling of personal information is described in our Privacy Policy. Where AI providers process personal information on our behalf, we seek to apply appropriate controls that are proportionate to the nature and sensitivity of the information and the proposed use.

Managing risk and maintaining human control

Before adopting a new or materially changed AI use, we consider relevant risks, including inaccurate or misleading outputs, bias, privacy, security, intellectual property, misuse and potential impacts on people.

Our controls are proportionate to the level of risk and may include:

• human review and validation;
• testing before production use;
• restrictions on the data that may be entered into AI tools;
• access controls and approved-tool requirements;
• monitoring for errors, changes in performance and emerging risks; and
• the ability for people to pause, override or discontinue an AI-supported process where appropriate.

People remain responsible for decisions and work products that they create, approve, share or use with AI assistance.

Transparency, feedback and concerns

We aim to be clear when AI is used in a way that directly engages with people, generates material content for them, or may meaningfully affect them.

If you have a question, concern or complaint about Larrison’s use of AI, please contact us at info@larrison.com.au. We will consider and respond to concerns in accordance with our applicable policies and processes.

Review of this statement

We review this statement at least annually and sooner if our AI use changes materially, a new AI-related risk is identified, or new information affects the accuracy of this statement.