Governing Intelligence · Version 0.2, July 2026

Govern AI the way your agency will be asked to answer for it.

A strategy, governance and assurance framework for artificial intelligence in Australian Government. Five domains held in balance as one system, anchored to the Commonwealth’s AI mandate, and built to be applied, not admired.

Take the free assessment Read the framework

Policy for the responsible use of AI in government v2.0
AI technical standard
National assurance framework
Guidance for AI Adoption (AI6)
PSPF and ISM
Privacy Act ADM transparency

What is due

The Policy for the responsible use of AI in government, and the dates around it

Checked against official sources on 7 October 2026. Binding on non-corporate Commonwealth entities.

15 June 2026A strategic position on AI adoption, communicated to staff
By July 2026A Chief AI Officer in every agency (APS AI Plan)
10 December 2026Automated decisions using personal information disclosed in privacy policies (Privacy Act)
15 December 2026Use-case owners, the register, impact assessments for new use cases, responsible AI practices and incident processes, and staff training
30 April 2027Existing use cases assessed and brought into line

The three tests

Every recommendation is held to three questions a citizen is entitled to ask.

Safe

The AI does what it is meant to and not what it is not. It is lawful, fair, contestable and tested against reality, and a person stays accountable for its outcomes.

Secure

The AI, its data and the systems it runs on are protected to Commonwealth standards. Data stays sovereign, access is least-privilege, third-party tools are contained.

Effective

The AI delivers measurable public value, with governance proportionate enough that safety and security do not smother the benefit.

The Governed Intelligence Model

Five domains. Two loops. One system to keep in balance.

Strategy sets the ambition and the risk appetite. The operating model fixes who is accountable. Architecture builds the secure, sovereign foundation. Design makes each system purposeful, safe and contestable. Governance assures it over time and feeds what it learns back into strategy.

An enablement loop carries the agency forward; an assurance loop runs the other way, so ambition never outruns the foundation and delivery never outruns safe practice. Most failures are one domain racing ahead of the others.

How the model works

StrategyAmbition and risk appetite
Operating ModelAccountability and value flow
ArchitectureSecure, sovereign foundation
DesignSafe, contestable systems
GovernanceAssurance and feedback

1AssistiveAI suggests; a person decides and acts on everything.
2AugmentativeAI drafts or acts on narrow tasks; a person approves each output.
The supervision threshold: assurance moves from reviewing outputs to governing the system as it runs
3SupervisedAI acts within bounds; a person supervises and can veto or roll back.
4AutonomousAI acts without per-action review, governed by policy and monitoring.
5OrchestratedMany AI agents coordinate as a system.

The AI Autonomy Ladder

Never grant AI more autonomy than your foundations can hold.

Each rung gives AI more decision rights, and each demands more of architecture, design and governance. Most government AI sits at rungs one and two today, and that is valuable, comparatively low-risk work.

The step from approving each output to supervising autonomous action is where promising pilots stall. Cross it deliberately, with the runtime controls in place, not under delivery pressure.

Matching ambition to readiness

Read the framework

Fourteen sections and twelve ready-to-use templates.

Leaders can start with the executive summary and the executive agenda. Architects, risk, privacy and security teams will find the operating detail in Part II. Delivery teams can begin with applying the framework, and the templates.

Part I. The case and the compass

1Executive summaryArtificial intelligence is moving government from automation to agency. For three decades, digital systems automated tasks while people held the...2Why AI governance, why nowThree forces have converged to make AI governance an immediate obligation rather than a future ambition: the nature of the technology has changed, the...3The Australian Government AI mandate landscapeAn agency cannot govern AI well without a clear view of what it is obliged to do. The difficulty is that the obligations are spread across more than a...4The North Star: the Governed Intelligence ModelTo govern AI across an enterprise, an agency needs more than a list of controls. It needs an organising idea that connects ambition to delivery,...

Part II. The five domains

5StrategyStrategy is the force that seeds everything else. It defines what the agency is trying to achieve with AI, how far and how fast it intends to go, and...6Operating ModelIf strategy is the spark, the operating model is the engine that turns it into coordinated movement. It defines who is accountable for AI, how...7ArchitectureArchitecture is the force that decides whether AI can scale safely or fails brittle. It is the foundation of data, systems and infrastructure on which...8DesignDesign is the domain that determines how an individual AI system actually behaves: what it perceives, what bounds it, what it optimises for, and how it...9GovernanceGovernance is the domain that makes the other four trustworthy over time. It provides the transparency, accountability and feedback that keep AI...

Part III. Putting it to work

10The AI assurance lifecycle and gatesThe five domains describe the capability an agency needs. The assurance lifecycle is how that capability is applied to each AI use case, consistently,...11The maturity journey and a roadmapAdopting this framework is itself a journey, and it should be paced by readiness rather than ambition. This section sets out where most agencies start,...12Leading governed AI: the executive agendaFrameworks do not govern; leaders do. Everything in Sections 5 to 11 can be built, and still fail, if AI governance is treated as a technical program...13Applying the framework in your agencyA framework earns its keep in contact with a particular agency: a particular size, mission, risk profile and starting point. This section answers the...14Recommendations and next stepsThis framework is designed to be adopted, not admired. The steps below convert it into action, and they align with the policy’s phased 2026 deadlines....

Appendices

A-LTemplates and appendicesUse this template to register a proposed AI use case and assign its risk tier. It produces a register entry and the first input to the impact...

Free assessment

Find out where you stand, and leave with a strategy.

Twenty questions across the five domains and your Policy v2.0 obligations, about fifteen minutes. You get a maturity score, the horizon to start from, your deadlines, and a tailored AI governance strategy and roadmap as a Word document your executive can edit. Nothing is sent to Renly: the whole assessment runs in your browser.

Start the assessment

Maturity across the five domains
Your obligations and dates, counted down
A ninety-day mobilisation plan
Strategy and roadmap, ready to edit in Word

From framework to runtime

The controls the framework asks for, running in Renly.

Autonomy gates become per-agent action rules: allowed, ask first, or never. Human participation becomes approvals bound to what a person reviewed. The decision passport becomes an Evidence Passport on every response, sealed into a hash chain so tampering is detectable, with one-click halts close at hand.

See how Renly maps to the framework

Autonomy gatesAction rules: allowed, ask first, never

Human participationApprovals bound to what the person reviewed

Decision passportSealed Evidence Passports

Kill switchHalt automations; pause computer use

Use-case registerAI Use Register with risk tiers and gates (Enterprise)

Questions, answered.

Who is this framework for?

Australian Government agencies that must govern and assure AI across the organisation, and the leaders, risk, privacy, security and delivery teams who will run it. It is written to the Commonwealth mandate, and state, territory and private-sector organisations that serve government can adopt it too.

Is it free to use?

Yes. Read it online, adopt it, adapt the templates and put your agency’s name to the resulting policy. The assessment is free as well, and your answers never leave your browser.

Is this legal advice?

No. It maps the obligations as they stood in mid-2026 and shows how to meet them. Confirm current obligations against the source instruments, which may have changed since this version.

How does it relate to the DTA policy?

It is built to discharge the Policy for the responsible use of AI in government (v2.0): strategy, accountable officials, the use-case register, impact assessments and transparency statements, aligned to the AI technical standard and the national assurance framework.

How does Renly fit in?

Renly is a governed AI runtime that implements the framework’s runtime controls: action rules that go ahead, ask a person or never act, approvals bound to what a person saw, sealed Evidence Passports, a tamper-evident audit trail and kill switches. The framework stands on its own; Renly is one way to run it.

Version 0.2, July 2026. First drafted by UHS Logic and published by Renly as a free, platform-neutral reference. Facts checked against official sources on 7 October 2026. Confirm obligations against the source instruments, which can change.