Artificial Intelligence & Automated Decision-Making Policy

1. Purpose and scope

This policy governs the responsible development, procurement, and use of artificial intelligence and automated decision-making systems by UniMatter, whether in the delivery of client engagements or in the operation of the firm. It reflects the firm’s commitment to method-led, accountable, and lawful practice, and is informed by Australia’s AI Ethics Principles and the firm’s obligations under the Privacy Act 1988 (Cth) and the Australian Consumer Law.

This policy applies to all artificial intelligence systems used by or on behalf of UniMatter, including machine-learning models, generative models, and rule-based automation, and to any process in which such a system materially informs a decision that affects an individual or a client.

2. Principles

UniMatter applies a consistent set of principles to its use of artificial intelligence. These principles align with Australia’s AI Ethics Principles and are applied proportionately to the risk presented by each use.

1. Human benefit and wellbeing: systems are used to deliver legitimate benefit and not to cause foreseeable harm.
2. Fairness: systems are assessed for bias and are not used to produce outcomes that unlawfully discriminate.
3. Privacy and security: systems handling personal information comply with the Australian Privacy Principles and the firm’s security policies.
4. Transparency and explainability: the role of automation in a decision is disclosed, and decisions are capable of meaningful explanation.
5. Contestability and human oversight: a person affected by an automated decision can seek review by a competent human.
6. Accountability: the firm remains accountable for the outcomes of the systems it uses.

3. Human oversight and accountability

UniMatter does not permit a consequential decision affecting an individual to be made by an automated system without meaningful human oversight. A competent person reviews and is accountable for outputs that materially inform such a decision, and retains the authority to override the system.

Automation supports professional judgement; it does not replace it. The firm remains accountable for the work it delivers irrespective of the tools used to produce it.

4. Data governance for AI

Personal information is used in the development, tuning, or operation of an artificial intelligence system only where there is a lawful basis to do so and the use is consistent with the purpose for which the information was collected. Client data is not used to train models for the benefit of other clients or third parties without authorisation.

Where third-party or generative artificial intelligence services are used, the firm verifies that client and personal information is not retained or used by the provider to train its models other than as authorised, and contracts on that basis under the Third-Party and Vendor Risk Policy.

5. Accuracy, validation, and limitations

Artificial intelligence outputs are validated against their intended purpose before they are relied upon. The firm tests systems for accuracy and reliability, monitors their performance over time, and records their known limitations.

Generative outputs are reviewed for factual accuracy, and any claim, figure, or citation produced by an automated system is verified against an authoritative source before it is relied upon or communicated. Illustrative figures are labelled as illustrative.

6. Transparency and contestability

Where an automated system materially informs a decision affecting an individual or a client, UniMatter discloses that fact and is able to explain, in meaningful terms, the basis on which the decision was reached.

An individual or client affected by such a decision may request human review. The firm provides a means to raise the request and ensures that the review is conducted by a person with the authority and competence to alter the outcome.

7. Risk assessment

Before a new artificial intelligence use is adopted, UniMatter assesses its risk, having regard to the sensitivity of the data involved, the consequences of error, the potential for bias, and the degree of autonomy afforded to the system. Higher-risk uses are subject to more stringent controls, oversight, and documentation.

Where a use involves personal information and is likely to have a significant effect on individuals, the firm conducts a privacy impact assessment.

8. Prohibited uses

UniMatter does not use artificial intelligence to make consequential decisions about individuals without human oversight, to engage in unlawful surveillance, to produce outputs that mislead or deceive contrary to the Australian Consumer Law, or in any manner that contravenes the firm’s legal obligations or this policy.

9. Review

This policy is reviewed at least annually and on any material change in the firm’s use of artificial intelligence or in the applicable legal and regulatory framework. Enquiries may be directed to UniMatter at administrator@unimatter.com.au.