BCS - Can AI Be Assured Without IT Sovereignty?

In this interactive evening, you will work in cross-functional teams to tackle a core framing question: Can an AI system be considered fully assured if critical elements of its operation remain subject to external legal, commercial or political control? Modern AI systems rely on complex ecosystems spanning multiple organisations and jurisdictions. A system may be demonstrably safe and compliant while still creating strategic dependencies that affect resilience, accountability and freedom of action. Working groups will examine whether sovereignty-related concerns—dependency, jurisdiction, concentration of control, verifiability and strategic autonomy—should be considered part of AI Assurance itself, or remain separate policy concerns. Team Activities: - Assessing safety risks from loss of critical suppliers, infrastructure, models or services - Evaluating security threats from foreign jurisdiction, supply-chain concentration and limited visibility - Identifying fairness concerns from concentration of data, compute and market power - Designing governance frameworks addressing control, portability, lock-in and evidence of ongoing operational control - Developing practical indicators for monitoring dependency risk and testing controllability The workshop will be supported with briefings, essential context and sources and a paper summarising the outputs will be published after the event.