

Pledge for Justice
A binding governance framework for artificial intelligence in the prosecution service, adopted before operational deployment. No decision to charge is automated. Every use of AI placed before the courts is disclosed. The framework and its training curriculum will be shared at no cost with other jurisdictions.
ThePledge
With a view to advance crime prevention and criminal justice, I pledge for the Republic of Maldives to adopt and bring into force, financed entirely from our own national budget, a governance framework for the use of artificial intelligence in our prosecution service, under which no decision to charge is automated and any use of artificial intelligence in material placed before the courts is disclosed, and to make that framework and its training curriculum available at no cost to other jurisdictions, to be implemented by the end of 2028.
Description of the Commitment
The Republic of Maldives commits to establishing a governed foundation for artificial intelligence in prosecution before, and not after, operational deployment. Through the Prosecutor General’s Office (PGO), the Maldives will adopt a binding governance framework for the use of artificial intelligence in the prosecution service, covering risk classification, human oversight, confidentiality, and disclosure to the courts. Every prosecutor and support officer will be trained and assessed as competent before using any such tool on a live case, and a register of tools in use will be maintained.
No decision to charge, to discontinue a charge, or otherwise to determine the liberty of a person will be automated. Live case data and the systems that process it will remain on sovereign infrastructure under Maldivian control. The programme giving effect to this commitment is financed in full from the national budget of the Republic of Maldives, without external or donor financing, as an expression of national ownership of justice sector reform.
Recognising that small island developing states and other small jurisdictions face a shared difficulty in adopting technologies designed elsewhere, in languages other than their own, and on infrastructure they do not control, the Maldives commits to making the governance framework, its supporting policy instruments, and the prosecutor training curriculum openly available at no cost to any interested jurisdiction — including the lessons of building Dhivehi language capability into justice sector systems. The Maldives stands ready to share this work with the UN Office on Drugs and Crime (UNODC) and with interested Member States.
Grounded in the UNESCO Recommendation on the Ethics of Artificial Intelligence (2021) and drawing on the UNESCO Guidelines for the Use of AI Systems in Courts and Tribunals.
Rationale and Timing
Artificial intelligence is being introduced into prosecution services, courts, and legal practice at a pace that has outstripped institutional preparation in most jurisdictions. For small States the material question concerns not the arrival of such technologies but the terms of their adoption: whether they are taken up under rules the State establishes for itself, or under conditions determined elsewhere and identified only once deficiencies in governance have become apparent in proceedings.
In the Maldives that choice remains open. Mafini, the case management system of the PGO, is currently being completed, and artificial intelligence capabilities are being incorporated within it rather than added subsequently. Where tools are already in service, a framework may regularise existing practice but cannot determine its design — the present cycle represents the point at which the commitment carries effect.
Three Features of the Maldivian Context
A small population
Information that would be unremarkable elsewhere may identify an individual. Privacy and re-identification risk require treatment at the design stage rather than reliance on general data protection provisions.
Dhivehi
The working language of prosecution in the Maldives is not supported by tools developed elsewhere. Optical character recognition and transcription in Dhivehi are treated as core requirements, benchmarked and validated before operational use.
Geographic dispersion
Prosecution services are delivered across an archipelago of more than 180 inhabited islands. Consistency of practice and equality of access require specific provision.
The programme timetable corresponds to the Congress cycle. Implementation runs to the end of 2028, with the governance framework in force in advance of that date and external review continuing thereafter. Progress is reportable through the Commission on Crime Prevention and Criminal Justice (CCPCJ) throughout the period 2026 to 2031. The framework is directed to governance, mandatory competency requirements, and published external review, rather than to any particular technology, and is designed for revision as the evidence base develops.
Expected Benefits
For the Criminal Justice System of the Maldives
Protection of rights and due process
Charging judgement remains with a qualified prosecutor. Where AI has materially affected material placed before a court, that use is disclosed and open to challenge.
Improved consistency and quality of casework
A target reduction of at least 40% in routine case preparation time by the end of the second year, together with fewer incomplete charging files.
Confidentiality and sovereign control
Live case data and the systems that process it remain on infrastructure under PGO control, reducing the risk of exposure through external platforms.
Service delivery in Dhivehi
Dhivehi capability, benchmarked before operational use, enables citizens throughout the atolls to engage with the prosecution service in their own language.
Systematic accountability records
An append-only ledger replaces the paper signature register, and a tamper-evident chain of custody applies to digital evidence.
Published external review
An external review of the effectiveness of the governance framework is published annually from the second year of operation, including measures that have not achieved their intended effect.
For Other Jurisdictions and the International Community
Most available guidance on artificial intelligence in justice has been developed by and for large, well-resourced jurisdictions. A framework developed within a small prosecution service, under constraints of budget, technical workforce, and language, is transferable in ways that such instruments frequently are not. Making the framework, the supporting policy instruments, and the training curriculum available at no cost provides other jurisdictions with an operational model rather than guidance alone — and the work on Dhivehi language capability is relevant to any State whose official language is not commercially supported. The annual external reviews are expected to contribute to the evidence base available to the CCPCJ during the Abu Dhabi cycle, including in respect of measures requiring correction.
Delivery and Implementation
The pledge is given effect by PGO 2.0, the institutional artificial intelligence strategy of the Prosecutor General’s Office. It is a 24-month programme delivered in four phases, financed in full from the national budget. Implementation follows a fixed sequence: adoption of the governance framework; training and assessment of staff; piloting of lower-risk capabilities; deployment of core capabilities; and thereafter the introduction of more advanced capability under strict oversight. No capability is used on a live matter before the framework is in force and the personnel concerned have been assessed as competent.
| Period | Principal Milestones |
|---|---|
| Months 1 to 6 | Governance framework adopted and risk classification in place. Training delivered across the PGO. Two capabilities piloted in controlled, non-live environments. Integration architecture approved. |
| Months 7 to 12 | Prosecutor assistant in supervised use. Case summarisation and charge drafting under human review. Evidence digitisation and Dhivehi and English transcription live for new matters. Document exchange ledger replaces the paper signature register. |
| Months 13 to 18 | Decision-support analytics in use on live matters. All capabilities integrated. Disaster recovery installed and tested. All staff assessed as competent. First external governance audit reported to the Prosecutor General. |
| Months 19 to 24 | Full deployment, capability transfer, and handover to PGO's own information technology function, against a defined handover standard. |
| To end 2028 | Framework in force and applied across all matters. External review published annually. Framework, policy instruments, and training curriculum available at no cost to other jurisdictions. |
Programme Governance
The programme is led by the Prosecutor General, with day-to-day management delegated to a programme director. A Designated AI Lead manages day-to-day AI governance, supported by an AI Governance Committee that meets at least quarterly. An independent technical advisory panel, including an external specialist in AI governance and a practitioner from a comparable jurisdiction, advises on technical decisions and reviews governance effectiveness at the close of each phase. Any tool may be suspended pending review and does not return to use until the matter is resolved.
Identified Risks
The 24-month timetable is demanding; scope may be re-sequenced at the close of a phase, without departure from the governance sequence. Capabilities dependent on other institutions, including the courts and the Maldives Police Service, are scoped early but implemented only after internal foundations have been validated. Recurrent operating costs following handover are provided for within PGO’s budget cycle from the third year, so the commitment is sustained beyond the construction period.
Maldives 2.0 and the Support of the Government
The Prosecutor General’s Office is a constitutionally independent institution, and the Prosecutor General is accountable for prosecutorial decisions. The commitment in this pledge rests on PGO’s own authority and its own funded programme, not on the budget or timetable of another institution, and may therefore be sustained across the full 2026 to 2031 cycle.
PGO 2.0 is delivered as an initiative under Maldives 2.0, the national digital transformation vision set out by His Excellency the President of the Republic in May 2025, under which the Maldives is committed to becoming a sovereign, inclusive, and digital-first State. The prosecution service is among the first institutions to give effect to that vision, and the first to do so in the justice sector.
The support of the Government is political and financial. Decisions on prosecution, and on the use of artificial intelligence in prosecution, remain those of PGO. No organ of government directs a charging decision or determines how artificial intelligence is applied in prosecution; that authority rests with the Prosecutor General and is unaffected by this pledge. Internationally, the commitment advances Sustainable Development Goal 16, in particular the targets on access to justice, accountable institutions, and public access to information.
Institutional Basis and Readiness
Four Factors Support Delivery
The approach draws on the UNESCO Recommendation on the Ethics of Artificial Intelligence (2021) as its normative foundation, the UNESCO Guidelines for the Use of AI Systems in Courts and Tribunals, guidance issued for prosecutors by the Prosecutors’ Center for Excellence (2025), the Guide for Using Generative AI in the Legal Sector issued by the Ministry of Law of Singapore (2026), the Practice Note on generative artificial intelligence issued by the Federal Court of Australia (2026), and the United Nations Guidelines on the Role of Prosecutors (1990). Where those instruments presuppose resources not available in the Maldives, the necessary adaptation has been made and recorded.
PGO also records the constraints affecting delivery: the national technical workforce is small, hardware lead times are extended, and construction depends on service providers. These are addressed through capability transfer as a contractual requirement, the embedding of PGO’s own information technology staff in delivery from the first phase, and a defined handover standard.
Categories, Timeframe and Reporting

