OFA Symposium 2025: Open Technology Impact in Uncertain Times

Wayne Wei Wang

Wayne Wei Wang is a Non-Resident Fellow at the FGV Center for Technology and Society (Brazil) and an ACCP Fellow at the African Center for Cyberlaw and Cybercrime Prevention (South Africa). He recently completed a Ph.D. in Law and Technology at the University of Hong Kong. With an interdisciplinary background in Engineering and Law, his research focuses on Intellectual Property, Data Protection, AI Governance, and Science & Technology Studies, with an emphasis on Law, Innovation, Sustainability, and Technology (LIST) in the Global South. His work appears in Computer Law & Security Review, Journal of AI Law & Regulation, and edited volumes from Oxford and Cambridge. Wayne holds an LLM from WIPO-QUT (Australia) and is an Associate Fellow of the UK Higher Education Academy. Wayne co-drafts the Chinese Model AI Law, deemed as one of the two emerging legislative proposals on AI in China. He received the 2025 PTC Emerging Scholar Award granted by Pacific Telecommunications Council and contributes to the UN IGF’s Dynamic Coalition on Data and AI Governance.


Session

11-18
14:50
30min
Post-DeepSeek AI Governance Debates: Open vs. Closed, Large vs. Tiny, Bytes vs. Watts, Horizontal vs. Vertical, Innovative vs. Safe?
Wayne Wei Wang

The release of DeepSeek V3 and its open-source inference model R1 marks a consequential inflection point in the global political economy of artificial intelligence. Achieving near-frontier performance at an order of magnitude lower cost and compute intensity than its U.S. and European counterparts, DeepSeek disrupts prevailing assumptions about the institutional prerequisites and technical economies of advanced model development, provoking challenges and criticisms such as model distillation. DeepSeek is not merely a technical episode — it is a prompt. Its trajectory invites a fundamental reconsideration of how openness, sovereignty, and interoperability operate under conditions of structural inequality, and how cost-efficient, legally navigable AI systems may establish new normative baselines for global technological ordering post DeepSeek.

This paper situates the “DeepSeek moment” within a broader reconfiguration of AI sovereignty, emerging from jurisdictions across the Global Majority — states in Asia, Africa, Latin America, and the Middle East—whose infrastructural agency and normative priorities remain peripheral within dominant transatlantic AI governance regimes. Employing a structural-institutionalist methodology, the paper analyzes how regulatory arbitrage, compute localization, and algorithmic adaptation converge in AI’s developmental arc. The model’s reliance on PTX-layer optimization, rather than full architectural overhaul, exemplifies an engineering strategy shaped less by maximalist innovation than by material constraint and jurisdictional heterogeneity. In doing so, it reframes openness not as a liberal universal but as a governance modality responsive to asymmetrical power. The paper explores five axes along which DeepSeek rearticulates global AI governance: (1) open versus proprietary development; (2) scale versus sufficiency; (3) data intensity versus energy efficiency; (4) vertical integration versus modular design; and (5) innovation versus precaution. Across these axes, DeepSeek displaces binary frameworks entrenched in geopolitics-driven regulatory imaginaries and signals the rise of institutional pluralism in AI normativity and infrastructure.

Open Source and AI
Main Room