Trustworthy AI for All

Summary

The “Trustworthy AI for All” project undertook a targeted three-month collaboration to address a critical gap in national AI strategies: the lack of a structured method for governments to make component-level decisions across the AI value chain. Existing strategies frequently invoke “AI sovereignty” but fail to specify where control is needed or how it interacts with global supply-chain dependencies.

Our core objectives were to:

• Define a comprehensive AI Stack Framework (spanning hardware and infrastructure, data, models, and applications)
• Review current global national AI policies to assess how control and dependency are currently distributed
• Develop a practical decision framework for countries to determine where to exercise control (Own, Co-own, Regulate, Procure, Depend) based on technical, institutional, and economic capacities
• Establish foundations for long-term international collaboration on trustworthy AI as a public good

Collaborating with multiple international research partners including; Singapore AI Safety Institute; Nanyang Technical University; Carnegie Mellon University; CMU Africa.

Activities and Achievements

The project was executed across three phases:

Phase 1 (January 2026): We conducted a structured comparative review of national AI strategies across Africa, the Americas, Europe, Asia, and Australia, and developed the initial AI Stack Framework.

Phase 2 (February 2026): We hosted a two-day international workshop in the UK, bringing together 20 experts from 13 partner institutions across five continents (including Carnegie Mellon, NTU Singapore, and EPFL). Participants refined and tested the framework, compared national AI strategies, and examined the role of AI in Digital Public Infrastructure (DPI).

Phase 3 (March 2026): We consolidated findings into practical governance tools, engaged policymakers and finalised the comprehensive project report and the accompanying policy brief.

The primary output is a detailed technical report titled “Democratising AI – Mechanisms to Develop Trustworthy National AI Capability,” accompanied by a targeted policy brief designed for government digital ministries.

Impact

The framework has immediate emerging impact by providing governments, particularly those with constrained resources, a mechanism to move away from binary “sovereignty vs dependency” debates. By disaggregating the AI ecosystem, we have equipped policymakers with the tools to manage supply-chain chokepoints (e.g., semiconductor concentration) and close the “enforcement gap” in current AI governance. The project successfully reframed national AI capability as a practical trade-off between cost, resilience, and responsible deployment.