PNG Government Artificial Intelligence Adoption Framework (DG × DPI ^ AI)

What it is

DG × DPI ^ AI is the Papua New Guinea Government’s draft framework for adopting AI in the public sector. The name sets out the logic. Digital Government (DG) and Digital Public Infrastructure (DPI) such as the SevisPass digital identity, SevisWallet and the SevisDEx data exchange must advance together, and AI is applied on top of those foundations to accelerate them rather than deployed as isolated tools.

The framework has two parts. Part 1, the Citizen Guide, was published for consultation on 27 February 2026 as version vPublic-1, with forewords from the Prime Minister and the Minister for ICT. It explains what government will and will not use AI for and what protections people have. Part 2, the Government Implementation Pack, holds the operational standards, safety checks, audits and rollout plans that agencies must follow.

The Acting ICT Minister’s press statement of 24 February 2026 said the draft provides for a governance and accountability framework, including consideration of an AI Taskforce, and calls for an AI legislative framework to support implementation.

Who it applies to

PNG Government departments and public bodies. DICT coordinates the national approach, and each public body stays responsible for its own services and decisions. The framework does not regulate private organisations, though vendors supplying AI to government will meet its safeguards in procurement.

What it requires

The Citizen Guide sets six red lines.

• No AI-only decisions on rights, benefits or penalties. Where important decisions occur, human review and accountability must apply.
• No secret citizen scoring. Government will not score or rank people in ways that affect rights or services without transparency and lawful authority.
• No use of personal data outside lawful purposes.
• No unsafe or untested AI in high-impact services. Use in sensitive areas must be introduced carefully and monitored.
• Extra care for culture and heritage. AI must not misuse PNG cultural heritage and expressions.
• No uncontrolled external AI in core government decisions. Unapproved external AI systems must not directly control or make core decisions without lawful authority, oversight and national safeguards.

It also sets out protections. Government commits to privacy and data protection, checking systems for bias, being clear where practical when AI supports a service, keeping public servants accountable for decisions, and giving people a way to ask for an explanation, request human review and complain. Section 5.7 commits government to keep internal registers of AI systems used in public services, to require systems to meet safety, fairness and security standards before deployment, and to assess systems by risk level, with stronger review, testing and monitoring for higher-impact systems.

Rollout is phased. It starts with lower-risk uses such as helping people find services and reducing form errors, expands where safeguards are proven and staff trained, and scales into high-impact areas only with extra controls. Government says it will publish progress reports covering safeguards, complaints and incidents.

For an agency, the practical work is a register of each AI system with its risk level and owner, an assessment before deployment, a documented human review point for decisions about people, and a way to receive and act on complaints.

Key dates

• 24 February 2026. The Acting Minister for ICT announced the draft framework.
• 27 February 2026. The draft Part 1 Citizen Guide was published for public consultation.
• 22 March 2026. The Ministry announced the draft National Sovereign Digital Transformation and AI Strategy, which builds on the same DG × DPI ^ AI model, for publication on 23 March 2026.

How it relates to other frameworks

The framework’s commitments on human review and appeal are written into law in the draft Data Governance and Data Protection Bill 2026, whose section 69 would bar public bodies from solely automated decisions on rights without meaningful human review. The draft National Sovereign Digital Transformation and AI Strategy goes further, proposing a National Artificial Intelligence Act and an independent National Digital Economy Authority that would classify AI systems by risk and set requirements for high-risk AI.

PNG’s 2024 data policy adopts the OECD AI Principles. The framework’s approach, with registers, risk tiers and human review, is close to New Zealand’s Public Service AI Framework, and agencies building the detail often use ISO/IEC 42001 as a management system structure.

A note on this page

This is Aicura’s reading of a draft government framework. It is not legal advice, and the framework may change before it is endorsed. Read the documents published by DICT, and check with DICT on how the final framework applies to your agency.