Ghana’s newly launched National Artificial Intelligence (AI) Strategy for 2025–2035 is facing early scrutiny from industry experts, with concerns emerging over internal inconsistencies that could complicate implementation of the country’s digital ambitions.
Desmond Israel, founder and lead consultant at Information Security Architects Ltd, said portions of the strategy suggest what he described as a “mechanical revision” of an earlier policy framework rather than a fully updated document aligned with its new timeline.
A central issue is the strategy’s coherence. While positioned as a 2025–2035 roadmap, sections of the document reportedly retain language referencing a 2033 endpoint, mirroring phrasing used in a prior 2023–2033 draft. That discrepancy, Isarael argues, goes beyond a typographical oversight and raises questions about how rigorously the document was updated.
“National strategies are built around timelines. If those timelines are inconsistent within the same document, it weakens confidence in execution planning,” he said.
The methodology underpinning the strategy has also drawn scrutiny. The document references stakeholder consultations conducted in 2022, without clearly distinguishing whether the latest version reflects new engagements or is primarily a revision of earlier work. According to Israel, such ambiguity risks obscuring the policy’s evidentiary basis.
“If it is a revision, it should be explicitly stated. If it incorporates fresh consultations, that process needs to be clearly documented,” he said, adding that transparency in policy development is critical for stakeholder trust.
Questions have also been raised about the institutional framework proposed to govern AI development. The strategy reportedly references multiple entities, including a Responsible AI Authority, a Responsible AI Office and a National AI Office, without clearly defining their respective mandates or how they would interact.
That lack of clarity could create confusion over accountability and oversight in a sector that spans regulation, public procurement, data governance and education policy.
“AI policy is governance infrastructure, not branding,” Israel said. “Without clearly defined institutions, implementation risks becoming fragmented.”
The critique comes as Ghana positions itself to expand its digital economy and integrate artificial intelligence into public services and industry. Policymakers have framed the strategy as a cornerstone for innovation, investment and skills development over the next decade.
Israel said the country’s ambitions remain achievable but warned that execution would depend on tightening the policy framework. “Where timelines change, update them consistently. Where institutions are proposed, define them clearly. Otherwise, avoidable defects can weaken implementation,” he said.
Another comment on the policy came separately from digital creator and cybersecurity expert Hector Dotse who said the more significant risk lies not only in inconsistencies but in what the strategy leaves unresolved.
“The bigger problem is what the strategy left out,” Dotse said, noting the document spans more than 80 pages and targets a GH¢ 500 billion contribution of AI to GDP by 2035.
He described elements of the strategy as “genuinely smart,” pointing in particular to its focus on language data. The plan includes a one-trillion-token target for Ghanaian languages by 2030, which Dotse said stands out as one of the document’s most compelling ideas. He also cited phased institutional design and what he called an “honest diagnostic” of the country’s current position.
However, Dotse said three critical decisions have been deferred. First is the funding architecture. While the strategy sets a GH¢200 billion private investment target, it does not outline the instruments required to mobilize that capital.
Second is the regulatory model. The document does not specify whether Ghana will adopt a risk-based framework similar to the European Union, a sector-led approach like the United Kingdom, a state-coordinated model akin to China, or sandbox-driven regulation as seen in Singapore.
Third is the technology stack. Dotse noted the absence of procurement principles ahead of major infrastructure commitments, including plans for a national AI compute centre.
“None of these requires reopening the strategy,” he said. “All of them require follow-on documents in the coming months.” He warned that the risk is not outright failure but misalignment over time.
