Published is not implemented: what six MENA AI strategies actually show
Five states in the region have published a national AI strategy. Fewer have named an authority to enforce one, and fewer still have an implementation programme beneath the vision. The gap between those three columns is where most national AI ambition is lost.
A national AI strategy is a cheap thing to announce and an expensive thing to deliver. The announcement earns a headline, a summit slot and a line in the country’s investment prospectus. The delivery earns none of those, which is roughly why the two diverge.
We compiled what six states in the region have actually published, with primary sources for every entry. The pattern that emerges is not about who is ahead. It is about which of three columns a state has filled in.
Three columns, not one
The first column is the strategy itself — a published document setting direction. Every one of the six has this.
The second is a named authority with a mandate over AI. This is where states begin to separate. Saudi Arabia has SDAIA, which functions as a standing institution rather than a committee. The UAE has both a dedicated Minister of State for Artificial Intelligence and a permanent AI Office. Egypt has a National Council for Artificial Intelligence. Where such a body exists, someone’s job depends on the strategy producing results.
The third column is an implementation programme sitting beneath the vision — the instrument that converts direction into sequenced, owned work. Oman is the instructive case here: its AI effort is framed as an Executive Programme under the National Digital Economy Programme rather than as a standalone vision document. That is the inverse of the common pattern, and it is the harder half done first.
Why the second and third columns predict everything
A strategy with no authority behind it has no mechanism for resolving the conflicts that implementation always produces. When two ministries disagree about who owns a dataset, a published vision statement cannot adjudicate; an authority with a mandate can. This is the single most common point at which national digital initiatives quietly stop.
A strategy with an authority but no implementation programme fails differently and more slowly. The authority exists, holds meetings, publishes principles — and cannot point to a departmental plan with dates and owners, because none was ever written. Eighteen months later the budget cycle arrives and there is no evidence to defend.
Saudi Arabia is worth noting for publishing both AI Ethics Principles and an AI Adoption Framework. Most states publish principles. Far fewer publish the instrument that tells an organisation what to actually do about them — which is the difference between a value and a control.
What this means if you are inside one of these systems
If your country appears in the first column only, the useful question is not “what should our strategy say.” It has been said. The question is who owns it, and what they are permitted to decide.
If you have an authority but no departmental plans, the gap is translation: nobody has written down what the national target requires of your directorate specifically. That is a matter of weeks of work, and until it exists nothing downstream can be planned, resourced or measured.
If you have all three, the remaining risk is cadence — the quiet decay of delivery pace between announcements. It is the most common failure mode and the cheapest to fix.
A note on what we did not claim
Nine states in our dataset are marked not yet researched. That means nobody has checked, not that no strategy exists. The distinction matters more than it might appear: in a field where confident assertion is abundant and citation is scarce, saying “we have not looked” is the more useful signal.