CAQ Artificial Intelligence Policy

Purpose and scope

1.1 Purpose. AI tools can save CAQ time on administration, writing and member communications. Used carelessly, the same tools can leak personal information, publish wrong facts under our name, or make decisions about people that no one can explain. This policy sets the small number of rules that let CAQ get the benefit without the harm.

1.2 Scope. This policy covers the CAQ management committee, CAQ staff, and volunteers acting for CAQ. It does not cover clubs or individual members. Guidance for clubs and members is a separate, later document. Where a club asks CAQ for advice in the meantime, we point them to the Australian Sports Commission and CSIRO guide, A Guide for Responsible AI in Sport.

Principles

CAQ's use of AI follows Australia's AI Ethics Principles, as adapted for sport in the ASC/CSIRO guide. In practice that means:

2.1 People decide, AI assists. AI never makes a final decision that affects a person's rights, membership or selection.

2.2 Privacy first. We collect and enter the minimum personal information needed, and only into tools that meet the tool-choice standard.

2.3 Honesty about AI. We say when AI made something.

2.4 Accuracy is ours. Whoever uses an AI tool owns checking its output. An AI error published by CAQ is CAQ's error.

2.5 Fair access. AI use must not disadvantage members who do not use, or cannot use, these tools.

Tool choice

Any AI tool that will touch personal information must hold a data processing agreement (or equivalent) meeting the European Union AI Act and GDPR standard. That means, at minimum: a written commitment not to train on our inputs, deletion rights, and documented data handling.

Paid or enterprise tiers of the major providers generally meet this bar. Free consumer tiers generally do not.

Personal information means member records, contact details, health or injury notes, and anything about a person under 18.

Never enter personal information about members, especially children, into free or public AI tools. This mirrors the ASC/CSIRO guide word for word and applies to everyone this policy covers, with no exceptions.

AI work with no personal information involved (general drafting, rules questions, research) needs no special tool approval. Use normal judgement.

Labelling

Content created by AI must be clearly labelled as AI-made. Labelled AI content is then judged on its value, the same as any other work, not dismissed because of how it was made.

Emails carry a stricter expectation. An email from a CAQ person is expected to be written by that person. Where AI wrote it, that must be made clear in the email itself.

Using AI as a tool inside your own work (checking grammar, tidying a table, summarising your own notes) is not AI-created content and needs no label. The test is authorship: if AI wrote it, say so.

Oversight and accountability

5.1 Human oversight. No decision that materially affects a person's rights or interests (membership, selection, eligibility, discipline) may be made by AI alone. From 10 December 2026 the Privacy Act requires organisations it covers to disclose automated decision-making in their privacy policy. CAQ writes to that standard now, whether or not the Act currently binds us.

5.2 Accountability. The accountable officer for this policy is the CAQ President. The accountable officer keeps a simple register of the AI tools CAQ uses and answers questions about this policy. Concerns about an AI tool or AI-made content go to the accountable officer first.

Incidents and review

6.1 If something goes wrong. Follow the existing CAQ escalation path, in line with the ASC/CSIRO guide. Raise it with the accountable officer first. Serious matters (a privacy breach, anything involving a child's data or image) escalate to the management committee, and where required to the external body: the Office of the Australian Information Commissioner for privacy breaches, the eSafety Commissioner for online harms, police where the law requires it. When in doubt, pause the AI use until the question is resolved. Record what happened and what changed as a result.

6.2 Review. The management committee reviews this policy annually, and sooner if the law changes. Two known changes to watch: the Privacy Act "Tranche 2" reforms, which would remove the small business exemption that currently keeps CAQ outside the Act (no date is set; the government has committed in principle only), and the EU AI Act's content-marking obligations, where AI content must be watermarked, in force from 2 December 2026, which will change what the major AI tools do by default.

References

[1] Department of Industry, Science and Resources (2025). Guidance for AI Adoption. Australian Government, October 2025. ai.gov.au.

[2] Department of Industry, Science and Resources (2024). Voluntary AI Safety Standard. Australian Government, September 2024.

[3] Department of Industry, Science and Resources. Australia's AI Ethics Principles. Australian Government.

[4] National AI Centre (2025). AI Policy Guide and Template, version 1.0. October 2025.

[5] Australian Sports Commission and CSIRO (2026). A Guide for Responsible AI in Sport. March 2026.

[6] Australian Sports Commission and CSIRO (2026). Roadmap for AI in Australian Sport.

[7] CSIRO. Artificial Intelligence Statement. csiro.au.

[8] Office of the Australian Information Commissioner. Privacy guidance for organisations and government agencies, including guidance for small business and sporting clubs. Made under the Privacy Act 1988 (Cth).

[9] European Union (2024). Regulation (EU) 2024/1689, Artificial Intelligence Act, Article 50, transparency obligations.

Colophon

Authored by CroquetClaude, 27 August 2026. Proofreading and very light editing by Wade Hart. Prepared for the Croquet Association of Queensland.

Status: accepted by the CAQ Management Committee on 22 September 2026.

Document control

This policy

Version history