Rethinking Data Sovereignty in the Age of Federated Learning
THE SCENARIO A consortium proposes federated learning as the solution to data sovereignty. The data never leaves the local node. Only model updates are shared. The privacy advocates celebrate. The sovereignty analysts notice: the model architecture is defined by the coordinator. The aggregation algorithm is controlled by the coordinator. The global model is owned by […]
The Trade Agreement That Became an AI Framework (2)
THE SCENARIO A digital trade chapter is negotiated as part of a broader economic agreement. The negotiators are trade specialists. The provisions cover source code, algorithms, data flows, and standards. The AI governance framework is written by trade lawyers who have never trained a model. The agreement enters into force. The sovereignty consequences are irreversible. […]
What Every Board Should Know About Model Weight Sovereignty
THE SCENARIO A corporation fine-tunes a foundation model on its proprietary data. The model weights are stored on the provider’s infrastructure. The licensing terms grant the provider a perpetual license to “improve” the model using customer interactions. The corporation’s competitive advantage — its data, its domain expertise, its institutional knowledge — has been legally transferred […]
The Hidden Cost of Algorithmic Non-Discrimination Clauses
THE SCENARIO A trade agreement includes an algorithmic non-discrimination clause. It sounds progressive. It prevents governments from favouring domestic AI models over foreign ones. The result: a national health service cannot prefer a locally trained diagnostic model because that would discriminate against the foreign incumbent. The clause protects competition. It eliminates sovereignty. When does a […]
Sovereign Cloud: Marketing Term or Governance Architecture?
THE SCENARIO Every major cloud provider now offers a “sovereign cloud” product. The marketing promises local data centres, local staff, local compliance. The contract reveals the control plane is managed from headquarters. The update pipeline comes from the global release train. The incident response escalates to a foreign security operations centre. The cloud is local. […]
The AI Workforce Crisis: Who Trains the Trainers?
THE SCENARIO A national AI strategy sets ambitious targets for domestic AI talent development. The universities expand their programmes. The training data comes from foreign platforms. The curved evaluation metrics come from foreign benchmarks. The certifications come from foreign vendors. The workforce is trained. The curriculum is not sovereign. What happens when your AI talent […]
Why Data Localisation Laws Are Failing Sovereignty
THE SCENARIO A government passes strict data localisation laws requiring all citizen data to remain within national borders. Cloud providers build local regions. Compliance boxes are ticked. Then a sovereignty audit reveals the encryption keys, the identity management, and the control plane all reside in a foreign jurisdiction. The data never left. The control already […]
The Sovereign’s Guide to AI Export Controls
THE SCENARIO A nation discovers its most advanced AI models fall under new export control regimes. The regulatory framework was written by the dominant AI-producing nations. Compliance means surrendering model weights to foreign auditors. Resistance means losing access to global compute markets. The sovereignty dilemma is absolute. What happens when your AI capabilities become someone […]