FRONTIER AI BEYOND BORDERS: RISK EXTERNALISATION AND GOVERNANCE GAPS IN NON-OECD STATES
Domaine:
digital infrastructureCréateur:
Olu
Éditeur:
Hôte:
The governance of frontier artificial intelligence remains almost entirely concentrated in OECD states, yet these systems operate as global infrastructure, with consequences that extend far beyond their jurisdictions of origin. This paper argues that the resulting asymmetry is not a temporary regulatory lag but a structural feature of the political economy of AI, one that systematically externalises risk to jurisdictions with lower institutional capacity and greater vulnerability to harm.
Through structured case analysis of four deployment contexts, namely Meta's content moderation failures during Ethiopia's Tigray conflict, Worldcoin's biometric data collection in Kenya, Indonesia, and Sudan, algorithmic credit discrimination in India, and AI-powered hiring bias in the Philippines, the paper identifies three compounding mechanisms of risk externalisation: training distribution bias that degrades performance in non-OECD contexts; contractual liability displacement that shields developers from downstream accountability; and regulatory capacity asymmetries that produce governance lags of 36 to 60 months in the jurisdictions least equipped to respond.
The paper makes three original contributions. First, it reframes frontier AI governance as a problem of global infrastructure governance rather than domestic technology regulation. Second, it demonstrates empirically how development-deployment asymmetries produce differential harm across institutional contexts. Third, it identifies a governance trilemma between maximising access, protecting vulnerable populations, and respecting sovereignty, which existing multilateral frameworks have not adequately confronted.
The analysis further shows how persistent governance gaps generate cloud infrastructure dependencies, enable regulatory arbitrage, and produce legitimacy deficits that threaten the coherence of emerging international AI governance architecture. The paper concludes that addressing deployment-side risk requires not only institutional capacity building but also deliberate political negotiation among actors with asymmetric power, and that continued inaction represents a governance choice with foreseeable and inequitable consequences.
Similaires
beyond-borders-org/TONAAI Governance Capacity in Developing Countries: Gaps, Innovations and ImplementationCode, Capacity, and Control: AI, Algorithmic Governance, and Political Accountability in African StatesBeyond borders: political marginalisation of Congolese young people in UgandaExploring AI governance in the Middle East and North Africa (MENA) region: gaps, efforts, and initiativesAI Governance in Nigerian Financial Services: An Assessment of Regulatory Frameworks, Institutional Gaps, and Reform Priorities