Executive Summary
Most global AI governance frameworks assume a process-based accountability model: if the procedure is documented and followed correctly, the decision is defensible. This assumption works in institutional-trust cultures where systems carry accountability. It fails — quietly but completely — in personal-accountability cultures where decision ownership must be traceable to a named individual.
In the GCC and many other regions, governance is not validated by process compliance. It is validated by the ability to answer three questions: Who approved this decision? Who can override it? Who is accountable if it goes wrong?
If no one owns the decision, the system is not governed — regardless of how well the process is documented. If the UN framework does not account for cultural governance models, it will be adopted on paper but bypassed in practice — leading to fragmented implementation, shadow systems, and ultimately, failure to scale.
1. The Problem Western Frameworks Assume Away
Current global AI governance standards are built on a foundational assumption:
If the process is correct, the decision is acceptable.
This works when institutional trust is high and accountability is distributed across systems. Audit the process. Document the steps. Follow the framework. The outcome is defensible because the institution absorbs responsibility.
But in personal-accountability cultures — common across the Middle East, parts of Asia, Africa, and Latin America — accountability does not rest with systems. It rests with people. Trust is personal. Responsibility is traceable. Governance is validated not by process documentation but by the ability to name the person who made the call.
When an AI system makes a recommendation and something goes wrong, the question is not was the process followed? The question is: who signed off on this?
If the answer is "the AI," the system is ungoverned — regardless of compliance documentation.
2. The GCC Case: When Governance Means Named Ownership
I have worked directly with government AI initiatives across Saudi Arabia — Ministry of Culture, SDAIA, and Vision 2030-aligned transformation programs. I have observed a consistent pattern:
When AI systems are deployed without clear human ownership of decisions, people do not reject the system loudly. They bypass it quietly. They override it informally. They build parallel decision processes alongside it. The system exists on paper. Decisions happen elsewhere.
This is not resistance to technology. This is a rational response to a governance gap. If an AI recommends a budget allocation, a resource deployment, or a policy decision — and no one can answer who owns that recommendation — trust never forms. Not because the AI is wrong, but because accountability is invisible.
The Three-Question Framework
Before any AI initiative is approved in a personal-accountability culture, these three questions must have clear, named answers:
Who approved this decision?
Not the process or the AI — a specific person with authority to commit.
Who can override it?
If the AI recommendation is wrong, who has the authority to stop execution before damage occurs?
Who is accountable if it goes wrong?
When scrutinised by oversight bodies, regulators, or leadership — who stands behind this decision and can explain why it was made?
These are not compliance questions. They are trust questions. And in contexts where trust is built on personal accountability, they are the only questions that determine whether an AI system will be genuinely adopted or quietly sidelined.
3. The Global Implication
The UN Global Dialogue has an opportunity to build a governance framework that works across contexts — not just in the environments where current standards were designed.
If the framework assumes process-based accountability as the default model, here is what will happen: countries will adopt the framework officially to demonstrate alignment with international standards. Organisations will implement the documentation and compliance requirements. AI systems will be deployed with full process adherence.
And those systems will be bypassed, informally, because the cultural accountability model was never addressed. The result is not rejection. It is fragmentation. The transformation program continues. The reports look healthy. But trust never forms, adoption never scales, and the investment never delivers. This is not hypothetical. I have observed this pattern repeatedly across government AI initiatives in the GCC. And the GCC is not alone — personal-accountability governance models are common across large parts of Asia, Africa, and Latin America.
4. Recommendation: Principle-Based, Not Process-Based
The UN framework should establish governance principles, not prescriptive processes. Those principles must include:
- Human Authority Over Outcomes. No AI system should be allowed to make a decision that no human is responsible for. This is the baseline, regardless of cultural context.
- Traceable Accountability. Every AI-informed decision must have a clear chain of human accountability that can be traced, audited, and explained to oversight bodies.
- Explainability Tied to Decision Authority. Explainability should not just describe what the AI did. It should clarify who approved it, who can override it, and who is accountable if it fails.
- Cultural Flexibility in Implementation. Countries should be free to implement these principles using governance models that align with their cultural accountability structures — whether institutional-trust or personal-accountability based.
This approach ensures global adoption without fragmentation. Principles are universal. Implementation is culturally grounded.
Conclusion
AI governance is not a technical problem. It is a trust problem. And trust is built differently across different cultures.
The UN has the opportunity to create a framework that works globally — not by imposing a single model, but by establishing principles that can be implemented in ways that align with how accountability actually functions in different contexts.
Are we building a framework that works on paper, or one that works in practice across the diversity of governance cultures that will need to implement it?
