Ghana's National AI Strategy 2025 to 2035: What It Means for Institutions Deploying AI

Ghana's National Artificial Intelligence Strategy, launched by President John Dramani Mahama on 24 April 2026, sets the terms on which AI will be adopted across the country's public institutions and priority sectors through 2035. It commits the state to eight pillars of action, a Responsible AI Authority, a National AI Office and a National AI Fund, alongside a Cabinet-approved $250 million national AI computing centre. This article sets out what the strategy contains, what it changes for the ministries, agencies, hospitals, universities and enterprises that will deploy AI under it, and how AdwumaTech, an AI services firm that designs, builds and deploys AI systems within institutions and governments across Africa, works within it.

The Strategy at a Glance

At the launch, President Mahama said that Ghana "will not be just a passive consumer of technologies shaping the future but that we are going to be an active participant in designing, governing and deploying them for our national transformation."

The Eight Pillars

1. Expand AI education
2. Empower youth for the jobs of the future
3. Deepen digital infrastructure and inclusion
4. Facilitate data access and governance
5. Coordinate a robust AI community ecosystem
6. Accelerate AI adoption across sectors
7. Invest in applied AI research
8. Promote AI adoption in the public sector

Four of the eight bear directly on institutions deploying AI systems: data access and governance, sector adoption, public sector adoption, and infrastructure and inclusion.

What the Strategy Changes for Institutions

The public sector becomes an AI buyer. Pillar 8 commits ministries, departments and agencies to adopting AI. Public procurement of AI systems will follow, and public buyers will need to justify each system to auditors, Parliament and the citizens it serves. The evidence a government AI system has to produce is set out in AI Assurance in the Public Sector.

Data governance moves to the centre. Pillar 4 and the Data Protection Commission's role in implementation place data protection inside every AI deployment. Institutions will be expected to show lawful basis, purpose limitation and security for the personal data an AI system processes, under the Data Protection Act 2012 (Act 843).

An oversight body is coming. The Responsible AI Authority is proposed as an independent body to oversee ethical AI deployment, monitor developments and coordinate implementation across sectors. Institutions deploying AI before it is established will be deploying systems the Authority later reviews, so records kept from the first release carry forward.

Sector adoption carries sector constraints. Pillar 6 names adoption across sectors, and each priority sector brings its own conditions. Health systems handle patient data under clinical oversight. Education systems operate at national scale across regions and languages. Agricultural services reach farmers through extension officers, voice and low-bandwidth channels.

Inclusion makes language a requirement. The strategy's human-centred and inclusive model, and Pillar 3's commitment to inclusion, mean citizen-facing systems serve people in the languages they speak. For a public service in Ghana, a system that operates only in English serves a fraction of the population it is accountable to.

Domestic compute changes where models can run. The $250 million computing centre adds in-country capacity for training and inference, which widens the options for institutions that need data and models to remain inside Ghana.

First-Year Actions for Institutions

• Inventory. List every AI system in use or procurement, with its purpose, data sources and owner.
• Accountability. Name a senior owner for AI across the institution and an operational owner for each system.
• Data protection. Complete a data protection impact assessment for each system that processes personal data.
• Alignment. Map each planned system to a strategy pillar and a priority sector, so funding and procurement cases rest on national priorities.
• Language coverage. Record which languages each citizen-facing system supports and which populations it leaves unserved.
• Records. Keep evaluation results, change logs and incident records from the first release, ready for review by the Responsible AI Authority.

The Deployment Readiness Score, a 12-question pre-deployment assessment, gives an institution a starting position against these actions.

Illustrative Workloads in the Priority Sectors

The following workloads are hypothetical and show how the strategy's priorities translate into systems.

Healthcare. A regional health directorate handles appointment, referral and results enquiries by phone in several languages. A voice system answers routine enquiries in the caller's language, escalates clinical questions to staff and writes a structured English record of each call to the directorate's system. Patient data stays inside the health service's infrastructure.

Education. A national examinations body receives candidate queries at peak volume around results release. A multilingual messaging system resolves status and procedure queries, routes exceptions to officers and records every interaction for audit.

Agriculture. An extension service reaches farmers by voice. A system transcribes and translates farmer questions from local languages, matches them to advisory content approved by extension officers, and aggregates the questions into a record of emerging issues by district.

How AdwumaTech Works Within the Strategy

AdwumaTech is an AI services firm headquartered in Accra that designs, builds and deploys AI systems within institutions and governments across Africa. Its work maps to six of the strategy's eight pillars, through an in-house engineering team in Accra, local-language AI, governed deployment inside the client's perimeter, published research, and academic partnerships.

Aligning the First System to National Priorities

AdwumaTech's AI Opportunity Assessment identifies the AI worth building in an institution and ties each candidate system to the strategy pillar and priority sector it serves. A ministry, hospital group or university receives a case for its first system that rests on national priorities, which is the basis on which public funding and procurement decisions will be made. Portfolio sequencing is covered in Which AI Use Case Should You Build First?.

Serving Citizens in Their Language

Nebula is AdwumaTech's platform for running AI applications across Twi, Ewe and Ga in its current release. Nebula provides speech recognition, machine translation and message routing, engages people in their language through voice and messaging channels, and produces an Interaction Record: a structured English record of each interaction, written to the institution's systems. For a public service, Nebula turns inclusion from a policy commitment into a measurable property of the channel: every interaction is served in the citizen's language and every interaction is recorded for the institution in English.

Governed Deployment Inside the Institution

AdwumaTech deploys systems on private cloud or local hardware inside the client's security perimeter, including state-controlled and domestic data centre infrastructure. Personal data, model weights and logs remain under the institution's control, which positions each system for the data protection obligations the Data Protection Commission enforces and for review by the Responsible AI Authority once it is established.

Evidence From the First Release

AdwumaTech delivers each system with its Evidence Chain record: impact assessment, data lineage, disaggregated evaluation, decision logs, oversight records and change control. The record answers the questions an oversight body, an auditor or Parliament will ask of a public AI system, and it is set out in full in AI Assurance in the Public Sector.

Research and Talent in Ghana

AdwumaTech's engineering team works from Accra, and its memoranda of understanding with the University of Ghana and Valley View University connect its engineering work to Ghanaian academic research. Its open speech datasets, including mGhana-ST, are published on Hugging Face for researchers building AI for Ghanaian languages, the applied research the strategy's seventh pillar calls for.

Related Resources

• AI Assurance in the Public Sector
• Operationalizing Enterprise AI in Ghana
• Introducing Nebula
• Deployment Readiness Score

About AdwumaTech

AdwumaTech AI designs, builds and deploys AI systems within institutions and governments across Africa. AdwumaTech is ISO/IEC 27001 certified, aligned to ISO/IEC 42001 and GDPR, and its mGhana-ST dataset won Best African Dataset at Deep Learning Indaba 2026.

Request an AI Opportunity Assessment