This page maps each requirement of the Australian Government’s Policy for the responsible use of AI in government against what Salsa Digital provides as a delivery partner. The policy is published by the Digital Transformation Agency (DTA), and this page refers to it as the DTA Policy v2.0. It took effect on 15 December 2025. The mapping gives an honest split of what Salsa Digital does and what your agency must still own.
This page covers our compliance support. For how we use artificial intelligence (AI) in general, see our AI transparency statement.
Who this is for
The DTA Policy is mandatory for all non-corporate Commonwealth entities (NCEs) and recommended for corporate Commonwealth entities. If your agency is in scope, you have obligations at both the agency level and the use case level. This page is for the people in your agency who are responsible for meeting them.
Salsa Digital isn’t an NCE and isn’t bound by the DTA Policy. We adopt it as our internal standard because most of our clients are bound by it. Doing so also makes our work clearly compatible with the rules our clients work under.
Our job is to provide delivery capability, technical controls, evidence and transparency that help your agency meet its obligations. Your agency still owns the policy, governance and accountability decisions. This page is an honest mapping of that split.
How we engage with you
Our responsibilities depend on what we’re doing for you. Three engagement modes come up across our work, and each has a different split of DTA Policy responsibility.
Mode A: We build AI capability for you
We design, configure and deliver an AI capability that becomes part of your systems. This includes AI features in a website we build, an integration we deliver, or a tool we configure for you. In this mode, Salsa Digital acts like a vendor for the AI capability. We provide technical controls, evidence and transparency for that capability. Your agency owns the use case under the DTA Policy.
Mode B: We operate or maintain AI in your environment
We run, manage or support AI already deployed in your systems. We share operational responsibility with you. We provide hands-on incident management, vendor change notification, monitoring signals and audit evidence. Your agency owns the use case and its governance decisions.
Mode C: We use AI internally on your work
We use approved AI tools inside Salsa Digital to help us deliver your engagement, such as drafting documents, reviewing code and summarising tickets. The AI isn’t yours, and you don’t need to register it as a use case in your agency’s register. But you have a right to know what we use, on what data and under what controls.
Across all three modes, our staff follow our internal AI policy and standard operating procedures, and work under our internal AI governance framework.
The responsibility split at a glance
Most of the DTA Policy applies to your agency. Our job is to make your compliance easier through delivery capability, controls and evidence.
Salsa Digital’s responsibility
- The AI capability we deliver and operate under Modes A and B (its design, configuration and ongoing operation)
- Salsa Digital’s own internal AI governance (Modes A, B and C)
- Mandatory annual AI training for Salsa Digital staff and contractors who use AI
- Technical controls on AI capability we build (data classification rules, output review, prompt sanitisation, approved tools only)
- Telling you within five business days when an approved AI vendor materially changes how it handles data, trains models, retains data or where it stores data
- Incident response for AI capability we operate
- Australian data residency for AI tools used on client work, with documented client acceptance for any exceptions
Shared responsibility
- Use case register entries for AI we built (we provide the inventory, you classify and register)
- Monitoring and re-validating in-scope use cases (we provide operational signals, you make governance decisions)
- Incident management (we manage what we operate, you own the response and any required notifications)
- Alignment with the AI Technical Standard for Australian Government
Your agency’s responsibility
- Your own AI transparency statement and strategic position
- Designating accountable officials and use case owners under your agency’s mandate
- Impact assessments and risk classification
- Mandatory staff training across your agency
- Pathways for staff and the public to report concerns
- DTA notification, six-monthly register submission and governance of high-risk use cases
Section 1: Strategy and oversight
| Requirement | Owner | What Salsa Digital provides | What your agency owns |
|---|---|---|---|
| 1.1 Public AI transparency statement published under the DTA standard, reviewed annually, DTA notified. | Agency | Our AI transparency statement is a vendor reference you can cite when you describe your use of AI that Salsa Digital delivers or operates in your own statement. | Draft and publish your agency’s statement describing your use cases, risks and mitigations. Review it every year. Notify the DTA at ai@dta.gov.au when you publish or update it. |
| 1.2 Strategic position on AI adoption published within 6 months of 15 December 2025. | Agency | This is your agency’s strategy. Salsa Digital doesn’t write it. You can engage us separately to advise on developing it. | Write your strategic position and communicate it to your staff. |
| 1.3 Accountable officials designated, DTA notified, under the Standard for accountability. | Agency | Salsa Digital publishes its Accountable Official for AI in its AI transparency statement. We don’t designate an Accountable Official for your agency. | Designate and document your agency’s Accountable Official, and notify the DTA. |
| 1.4 Accountable use case owners for each in-scope use case, within 12 months. | Agency, with shared input for Modes A and B | For Mode A engagements, Salsa Digital nominates an accountable use case owner inside Salsa Digital for the AI capability we deliver. This person works with your agency’s accountable use case owner. The two roles are different. Our owner is accountable for the supporting use case in Salsa Digital’s internal register. Your agency’s owner is accountable for your use case under the DTA Policy. For Mode B, we work with your designated owner. For Mode C, the use case sits in our own internal register and Salsa Digital owns it. | Designate the accountable use case owner in your agency for every in-scope use case, whether Salsa Digital delivered it or not. Your agency’s owner is the one accountable under the DTA Policy. |
| 1.5 Internal AI use case register created within 12 months, shared with the DTA every 6 months. | Shared | For Mode A engagements, Salsa Digital provides the inventory and metadata for the AI capability we delivered: agents, tools, integrations, configurations and the data they handle. You can use this directly in your agency’s register. Salsa Digital keeps a separate internal register of the five internal use cases summarised in our AI transparency statement. | Maintain your agency’s use case register in whatever system you use for governance. Classify each use case, designate accountable owners, and format and submit the register to the DTA every 6 months. |
Section 2: Preparedness and operations
| Requirement | Owner | What Salsa Digital provides | What your agency owns |
|---|---|---|---|
| 2.1a Process for adopting AI use cases aligned to risk and integrated with enterprise risk management. | Agency, with technical controls from Salsa Digital for Modes A and B | For Modes A and B, we apply our internal AI use case process as part of design and delivery. This includes screening with the National AI Centre (NAIC) tool, DTA in-scope assessment, impact assessment and risk treatment design. We document these as engagement records you can add to your governance. | Design and enforce your agency’s process. You can engage Salsa Digital to advise on it or run parts of it. |
| 2.1b Inform staff of Australia’s AI Ethics Principles. | Agency | You can engage us to develop or deliver training, but communicating the principles to your staff is your responsibility. Our own staff learn the principles in our mandatory annual training. | Communicate the principles to your staff. |
| 2.1c Staff pathway to report AI safety concerns and incidents. | Agency, with optional support from Salsa Digital | Your agency’s existing internal reporting channel (ticketing, email, line manager) is the right pathway. For Mode A and B engagements where Salsa Digital operates the AI, we provide audit records and operational evidence on request to support your triage. | Provide and operate the channel. Triage and fix issues. Escalate to accountable officials where needed. |
| 2.1d Public pathway to report concerns suited to your agency’s AI use. | Agency | The public-facing channel belongs to your agency. For Mode A engagements where Salsa Digital delivered a public-facing AI capability, we support your triage with audit evidence. To report concerns about Salsa Digital’s own AI use, the public can email security@salsa.digital. | Provide a public web form, email address or other channel. Route reports to the right team. |
| 2.1e Incident management aligned to information and communications technology (ICT) incident management and overseen by a governance body or senior executive. | Shared | For Mode A and B engagements, Salsa Digital runs incident management for the AI capability we operate, including detection, triage, containment and remediation. We notify you of incidents that affect your engagement within agreed timeframes. Separately, we notify you within five business days if an approved AI vendor we rely on materially changes how it handles data, trains models, retains data or where it stores data, in a way that affects the controls we’ve documented for your engagement. | Your incident response plan. Severity classification at the agency level. Post-incident review. DTA notification where it applies. |
| 2.2 Mandatory staff training on AI within 12 months. | Agency for your staff, Salsa Digital for ours | Our staff and contractors complete mandatory annual AI training before they use AI on any Salsa Digital work, including work for your agency. Training covers our internal AI governance framework, AI policy and procedures, data classification rules, prompt and output handling, incident reporting and Australia’s AI Ethics Principles. Completion records are available on request. | Deliver training to your own staff. Track completion. Refresh it regularly. |
| 2.3 AI Technical Standard for Australian Government strongly recommended. | Shared | For Mode A engagements, Salsa Digital applies the technical standard to AI capability we deliver. We design fairness, transparency, accountability, human oversight and contestability into the capability, and document them in the AI Services Security Assessment for that engagement. | Apply the standard to your use cases overall. Check the mapping fits your risk profile. |
| 2.4 AI procurement guidance strongly recommended. | Agency | We respond to standard agency procurement questionnaires. They cover data residency, retention, model training, breach process, vendor certifications and Salsa Digital’s own AI governance. | Run procurement in line with DTA guidance. Evaluate responses against your criteria. |
Section 3: AI use case impact assessment
| Requirement | Owner | What Salsa Digital provides | What your agency owns |
|---|---|---|---|
| 3.1 Assess all new use cases against the in-scope criteria. Existing use cases assessed by 30 April 2027. | Agency, with technical input from Salsa Digital for Modes A and B | For Mode A engagements, Salsa Digital prepares an in-scope assessment during the design phase and gives it to your agency to review and adopt. The judgement is yours. The input is ours. | Document the in-scope assessment at the design stage. Record the outcome in your register. |
| 3.2 Impact assessment for in-scope use cases using the DTA tool or an equivalent internal process, finalised before deployment. | Agency, with delivery support from Salsa Digital | For Mode A and B engagements, Salsa Digital can run the impact assessment as a contracted activity. We provide the technical content on ethics, privacy, security, fairness and stakeholder exposure. The assessment document is yours and lives in your governance or records system. We design risk treatments into the AI capability we deliver. | Finalise the impact assessment. Store it in your governance or records system. Apply risk treatments. Sign off before deployment. |
| 3.3 Register, monitor and re-validate in-scope use cases over time. | Shared | For Mode A and B engagements, Salsa Digital provides operational signals: incident reports, anomaly observations, audit records, usage metrics and notice of material changes to underlying tools. A material change in our tools triggers vendor change notification, as set out in 2.1e. | Set your review cadence. Act on signals. Revise impact assessments when assumptions change. Reclassify if scope changes. |
| 3.4 Medium-risk use cases consider board or senior executive governance. | Agency | We supply the operational and audit data your governance body needs for its review. | Decide whether extra governance applies, and run it. |
| 3.5 High-risk use cases mandatory board or executive governance, report to the DTA, review at least every year. | Agency | Evidence and operational data your accountable officials use for the annual review and DTA report. For high-risk Mode A engagements, Salsa Digital supports the annual review with a written summary of how the AI capability operated during the period. | Governance body oversight. Compile and submit the DTA report. Run the review at least every year. |
| 3.6 Out-of-scope use cases comply with privacy, security and other obligations, and are reassessed if scope or use changes. | Agency | All Salsa Digital controls (data classification rules, output review, incident reporting) apply to AI capability we operate, whether or not the use case is in scope under the DTA Policy. This gives a consistent safety net. | Reassess when scope, use or operation materially changes. |
How we help with the in-scope criteria
The DTA Policy’s five in-scope criteria (in Appendix C of the policy) decide whether a use case needs a full impact assessment. The classification is your agency’s judgement. Our controls help once you’ve made the call.
Personal or sensitive data (Privacy Act 1988)
We apply data classification rules across all our work. Personal information may only go into AI tools specifically authorised for that data. Cross-border processing of personal information is subject to Australian Privacy Principle (APP) 8, and Salsa Digital’s Privacy Officer reviews it before approval. Where approved tools process personal information, we document the arrangements for each use case. Our default is to use Australian-resident AI platforms for client-facing work.
Security-classified information (PSPF)
We apply the controls in Protective Security Policy Framework (PSPF) Policy Advisory 001-2025 when working with OFFICIAL information. Higher classifications aren’t allowed on AI tools without specific accreditation and explicit written direction from the client. Our internal data classification rules set what our staff may do.
Materially influences administrative decisions
Salsa Digital doesn’t use AI to make decisions on behalf of clients or members of the public. AI helps with analysis, drafting and summarising. People who are accountable for the decisions make them. This is a firm rule in our internal AI governance framework and AI policy.
Direct public interaction without human review
We won’t deploy public-facing AI for a client without your written approval, an impact assessment, and arrangements for human oversight and contestability that meet the DTA Policy’s expectations. Salsa Digital’s own internal AI use doesn’t interact with the public directly.
Potential for harm
We treat AI-related incidents, including near misses, as a type of security and operational incident. Our AI governance committee, chaired by our Accountable Official, reviews material incidents and oversees how they’re fixed.
Evidence we can provide on request
For a client engagement or a procurement process, Salsa Digital can provide:
- Our AI governance framework, AI policy and AI standard operating procedures
- A summary of our internal AI use cases
- Data residency, retention, vendor data handling and certification status for each approved AI tool relevant to your engagement
- Records of AI training completion for Salsa Digital staff and contractors (anonymised totals or, where a contract requires it, named individuals)
- The AI Services Security Assessment for your engagement, a client-facing document with tool-by-tool detail
- Responses to your agency’s standard AI procurement questionnaires
- Incident reports and audit records for AI activity Salsa Digital performed on your engagement, subject to any contract limits
- Operational evidence to support your impact assessment, register entries or annual review of high-risk use cases
Send requests to security@salsa.digital. Some documents may need a signed non-disclosure agreement (NDA).
Related documents and contact
- Salsa Digital AI transparency statement (how we use AI in general)
- Policy for the responsible use of AI in government (DTA Policy v2.0)
- DTA Standard for AI transparency statements
- DTA Standard for accountability
- AI Technical Standard for Australian Government
- Guidance on AI procurement in government
- NAIC AI screening tool
- Australia’s AI Ethics Principles
- Voluntary AI Safety Standard
- Australian Government Protective Security Policy Framework
- Australian Privacy Principles
- PSPF Policy Advisory 001-2025
Send questions about this page or any other part of Salsa Digital’s AI compliance to security@salsa.digital.