Artificial Intelligence (AI) Usage Policy – Steady Guide

💡 At a Glance & Governance Overview

⏱️ Estimated Read Time: 6 minutes

Central Takeaway: Steady Guide maintains an ethical, human-centric AI usage policy that enforces zero input of participant data, Australian data sovereignty, strict human oversight (Human-in-the-Loop), and complete physical segregation between AI tools and participant clinical systems.

• Policy Administration & Version Control: Version 2.0 updated 11 September 2026 under Jenny Mathers’ approval authority.
• Purpose, Scope, & Definitions: Aligned with the Privacy Act 1988, NDIS Code of Conduct, and NDIS Practice Standards.
• Authorised AI Technologies & Use Cases: Approved tools restricted to non-confidential research, drafting, and generic graphics.
• Mandatory Human Oversight & Approval: Strict Human-in-the-Loop (HITL) system requiring 100% review by Jenny Mathers.
• ‘Red Line’ Prohibitions: Absolute ban on entering participant data into AI models and zero automated decision-making.
• APP Compliance Framework: Full operational mapping across APPs 1, 3, 7, and 13\.
• Automated Decision-Making & Future Transparency: December 2026 disclosure mandate and right to human review.
• Infrastructure, Security, & Audit: Complete physical segregation of Astalty/M365 participant data from AI processing layers.
• Policy Review & Acknowledgment: Mandatory annual review and strict compliance enforcement under NDIS Code of Conduct.

Section 1: Policy Administration and Version Control

In the high-velocity landscape of National Disability Insurance Scheme (NDIS) regulation and the concurrent evolution of generative technologies, this policy serves as a cornerstone of organisational integrity. Maintaining rigorous version control is not merely an administrative task; it is a critical strategic mitigation against “regulatory drift.” By ensuring this “living document” reflects contemporary NDIS Quality and Safeguards Commission expectations and emerging Information Security (InfoSec) standards, Steady Guide maintains a state of continuous audit readiness. This administrative rigour provides the essential technical assurance required to protect participant safety and maintain institutional trust.

Formal Statement of Protocol

To ensure corporate transparency and high-integrity governance, all Steady Guide policies must henceforth include a standardised version control and update date block. This protocol facilitates a clear audit trail of policy evolution and ensures all personnel, contractors, and stakeholders are operating under the most current regulatory guidance.

Section 2: Purpose, Scope, and Definitions

Steady Guide is anchored by a commitment to “People, Potential, and Possibilities.” As a specialist provider of support coordination and case management, we recognise that the strategic adoption of Artificial Intelligence (AI) is essential for operational efficiency. However, technological utility must never supersede participant trust. This policy provides the governing framework for the ethical and legal integration of AI into our workflows, ensuring that technology serves to enhance human empathy rather than replace it.

Section 3: Authorised AI Technologies and Operational Use Cases

Steady Guide has executed a strategic shift in its information architecture, moving away from previous legacy environments to enhance data provenance and research integrity.

Operational Shift and Legacy Data Handling

Steady Guide uses Google Notebooks (Gemini Notebook) as its primary enterprise solution for research synthesis.

Approved Functions

The use of authorised AI tools is restricted to the following assistive, non-confidential activities:

• Research & Information Synthesis: Leveraging AI to synthesise disability sector trends, legislative summaries, and broad industry research.
• Content Drafting: Generating initial conceptual frameworks for educational resources, newsletters (e.g., Steady Steps), and internal policy templates.
• Visual Assets: Generating or altering generic marketing imagery for the website and social media.

These tools are strictly assistive. They function as a drafting aid and are subject to mandatory human gatekeeping before any transition to the public or operational domain.

Section 4: Mandatory Human Oversight and Approval Workflow

To mitigate the risks of “algorithmic hallucinations”—where AI generates factually incorrect but plausible-sounding data—Steady Guide mandates a strict Human-in-the-Loop (HITL) system. Google Notebooks are used primarily for its ability to reference only reputable sources chosen by Jenny Mathers. Information is cross-checked and fact-checked through these reputable sources.

Final Review Mandate

All AI-assisted outputs must be rigorously reviewed, edited, and approved by Jenny Mathers prior to publication or implementation. No content generated by AI is permitted to bypass this human verification stage.

Professional Judgment and Legal Precedent

Consistent with the principles established in the Ortiz v NDIA / Butler matters, Steady Guide acknowledges that AI outputs are not evidence on their own. They can be incomplete, biased, or incorrect. All AI-assisted findings must be critically validated against primary records, official NDIS Rules, and Commission guidance. AI is a tool for synthesis; it is never a substitute for professional judgment or duty of care.

Section 5: The ‘Red Line’ Prohibitions—Data Privacy and Integrity

These “Red Lines” represent the non-negotiable ethical and legal boundaries of Steady Guide. They are designed to protect the highly sensitive information inherent to NDIS service delivery.

Strict Prohibitions:

Zero-Input Mandate: There is an absolute prohibition against inputting participant personal or sensitive information into AI models. This is a critical defense against the ingestion of NDIS data into global Large Language Model (LLM) training sets, which would constitute a permanent breach of data sovereignty and participant privacy.

No Automated Decisions: AI must never be used to make solitary, unreviewed decisions regarding participant services, funding eligibility, or access to supports.

Data Residency and Sovereignty

To satisfy APP 8 (Cross-border disclosure), Steady Guide enforces the following:

• Australian Sovereignty: All AI tool vendors must provide written confirmation that data processed within their enterprise environment resides in Australian-based data centres. The data kept in Google Notebooks are not of a personal or confidential nature, however according to Google, this information is stored in data centres located in the same country as the user (Jenny Mathers).
• Training Prohibitions: Vendor terms of service must explicitly state that corporate data is not utilised for underlying model training.

Section 6: Australian Privacy Principles (APP) Compliance Framework

Steady Guide utilises this policy as its transparent management plan for AI-related data handling, directly supporting the Information Management outcome of the NDIS Practice Standards.

• APP 1 (Open Management): This policy is the transparent roadmap for how AI is used within our governance framework.
• APP 3 (Collection): Any data collected via AI interaction must be “reasonably necessary” for our current functions. We do not engage in speculative data collection for future use.
• APP 7 (Direct Marketing): AI-generated marketing copy must include a clear, prominent, and simple “opt-out” mechanism. Steady Guide will respect all unsubscribe requests within a reasonable period, typically not exceeding 30 days.
• APP 13 (Correction): We maintain a 30-day window to respond to requests for the correction of inaccurate AI-generated information. Pursuant to APP 13.2, if Steady Guide corrects personal information that was previously disclosed to another APP entity (such as a provider or the NDIA), we will take reasonable steps to notify that third party of the correction upon the individual’s request. Where a correction is refused, a written notice including the reasons and complaint process will be provided.

Section 7: Automated Decision-Making (ADM) & Future Transparency

Steady Guide takes a proactive stance regarding upcoming transparency requirements for automated systems.

2026 Disclosure Requirement

By December 2026, Steady Guide will update its Privacy Policy to include specific disclosures regarding what personal information is used in AI-assisted processes and the nature of the decisions these tools support.

Right to Human Review and Explainability

Every participant retains the right to request a human review of any decision meaningfully supported by software. This includes the right to an explanation of the AI’s role in the process. Steady Guide ensures that all technology-assisted decisions remain explainable, fair, and subject to human accountability.

Section 8: Infrastructure, Security, and Audit

The primary technical defense against data leakage is the physical and technical segregation of Steady Guide’s information environments.

Technical Architecture

• Participant Systems (Secured): All participant records, emails, and data are hosted within the Microsoft and Astalty environments. These utilise Australian-based data centres with bank-grade encryption and enterprise security.
• AI Assistive Tools (Segregated): Google Notebooks/Gemini operate in a separate, non-integrated environment. This physical separation ensures participant data is never exposed to the AI processing layer.

Case-Note Annotation

Where AI has been utilised to assist in drafting internal summaries or documentation, case notes must be annotated to reflect AI involvement. This ensures a transparent audit trail for internal review and Commission audits.

Spelling and Style Standards

All documentation must align with the Steady Guide Master Style Manual:

• Typography: All text must be strictly left-aligned (no justified text).
• Spelling: Use Australian Commonwealth spelling exclusively (e.g., “-ise” suffixes and “acknowledgment” without the middle “e”).

Section 9: Policy Review and Acknowledgment

This policy is a cornerstone of Steady Guide’s commitment to ethical service delivery and regulatory excellence.

Annual Review Mandate

This policy will be reviewed at least annually, or immediately upon significant regulatory or technological changes, to ensure continued alignment with NDIS Practice Standards and Australian law.

Final Acknowledgment

All personnel, contractors, and representatives must read, understand, and comply with these standards. Failure to adhere to the “Red Line” prohibitions is considered a serious breach of the NDIS Code of Conduct and may result in disciplinary action and notification to the NDIS Quality and Safeguards Commission.