Skip to content

AI Project Register

Purpose: Central record of all AI initiatives within your organisation for governance oversight Audience: PMO, ICT, risk and governance teams | Time: 30 minutes setup, ongoing updates

This register helps you maintain a central record of all AI initiatives. It:

  • Ensure visibility across all AI-related projects
  • Provide a single source of truth for governance, risk and compliance monitoring
  • Support decision-making through consistent project documentation and guardrail alignment

When to Use

  • 🎯 At project initiation: Create a new entry for each AI initiative
  • 🔄 During project lifecycle: Update details as the project evolves (e.g., risks, model versions)
  • ✅ At review points: Use the register to assess go/no-go criteria and ensure guardrail compliance

Relevant Guardrails: 1, 2, 9, 10 (from the Australian Voluntary AI Safety Standard)

AI Project Register Template (Template)

Project Register Fields

Section Field Description Example Entry
Project Information Project Name Title of the AI initiative Customer Insights Chatbot
Description Short summary of the project’s purpose Automating first-line customer queries using an LLM
Objectives Key business goals or outcomes expected Reduce response time by 40%
Timeline Planned start/end dates, key milestones Start: Aug 2025, Pilot: Nov 2025
Ownership & Governance Project Owner Person accountable for delivery Jane Smith, Head of CX
Stakeholders Business units and key contacts IT, Risk, Legal, Operations
Approval Status Formal governance decision Approved by ICT Steering Committee
Risk Assessment Guardrail Compliance Alignment with AI guardrails (Yes/Partial/No) Guardrails 1, 2, 9 compliant; 10 pending
Risk Level Overall risk rating (Low/Med/High) Medium
Mitigations Key risk controls applied Human-in-the-loop escalation for safety checks
Technical Details Data Sources Internal/external data powering the model CRM data, anonymised chat logs
Model Information Model type, vendor, or custom build details GPT-4o, fine-tuned
Infrastructure Hosting, deployment environment Azure Cloud, containerised
Financial Budget Allocated Total approved budget $250,000
Actual Spend Current expenditure $125,000
ROI Target Expected return 40% efficiency gain
Dependencies Related Projects Other initiatives this depends on Data Lake Project
System Integrations Systems this connects with CRM, ERP, Analytics
Lifecycle Pilot Date When pilot begins 1 Nov 2025
Production Date Go-live target 1 Feb 2026
Review Date Next formal review 1 May 2026
Sunset Date Planned decommission 1 Feb 2028
Ethics Ethics Review Status of ethical assessment Completed - Low Risk
Bias Testing Results of bias evaluation Passed all criteria
Benefits Benefits Realised Actual vs planned benefits 35% efficiency (target 40%)
Monitoring & Updates Version History Track model releases or changes v1.0 (Aug 2025), v1.1 (Oct 2025)
Performance Metrics Agreed KPIs or benchmarks Accuracy >85%, CSAT >90%
Change Log Notes of updates, retraining, risks Retrained with new dataset Sep 2025
Decision Framework Go/No-Go Criteria Conditions for continuation Meets KPIs, passes compliance review
Escalation Path Who is notified if risks emerge Escalate to CIO and AI Risk Committee

How to Maintain the Register

  • Ownership: The AI Project Register should be owned by the PMO, ICT, or Risk/Governance function.
  • Frequency of Updates: At minimum, quarterly updates, or more frequently for high-risk/high-impact projects.
  • Integration: Link the register with project governance forums, risk registers and compliance reporting.
  • Audit & Oversight: The register can be used as part of routine compliance checks to ensure responsible AI deployment.

Status Tracking

Overall Status: [ ] On Track [ ] At Risk [ ] Delayed [ ] On Hold [ ] Cancelled

Health Indicators:

  • Schedule: 🟢 Green / 🟡 Amber / 🔴 Red
  • Budget: 🟢 Green / 🟡 Amber / 🔴 Red
  • Risk: 🟢 Green / 🟡 Amber / 🔴 Red
  • Compliance: 🟢 Green / 🟡 Amber / 🔴 Red
  • Risk Assessment: [Document ID/Link]
  • Vendor Evaluation: [Document ID/Link]
  • Incident Reports: [Document ID/Link]
  • Ethics Review: [Document ID/Link]
  • Business Case: [Document ID/Link]

Alignment with Australian Standards

Standards Compliance

✓ Decide who is accountable — "Ownership & Governance" section requires "Project Owner" and "Stakeholders" for every initiative

✓ Understand impacts and plan accordingly — "Risk Assessment" section requires documentation of "Risk Level" and "Mitigations" before proceeding

✓ Share essential information — Register acts as central source of truth, sharing project details (Objectives, Timeline, Status) across the organisation

✓ Guardrail 1 – Accountability — "Project Owner" field ensures specific individual accountable for AI system outputs and impacts

✓ Guardrail 10 – Stakeholder engagement, safety, diversity, fairness — "Ethics" section (Ethics Review, Bias Testing) verifies fairness and safety

✓ Guardrail 9 – Record-keeping — Register implements record-keeping by maintaining history of all AI projects, status and key decisions

✓ Guardrail 2 – Risk management — "Risk Assessment" fields (Risk Level, Mitigations) integrate risk management into project lifecycle

Next Steps

Where to go from here:

Disclaimer & Licence

Disclaimer: This template provides best practice guidance for Australian organisations. SafeAI-Aus has exercised care in preparation but does not guarantee accuracy, reliability, or completeness. Organisations should adapt to their specific context and may wish to seek advice from legal, governance, or compliance professionals before formal adoption.

Licence: Licensed under Creative Commons Attribution 4.0 (CC BY 4.0). You are free to copy, adapt and redistribute with attribution: "Source: SafeAI-Aus (safeaiaus.org)"