What happened
Drawing on ten public discussions from the India AI Impact Summit 2026, Carnegie India analyzes how the open versus closed AI debate is evolving in the context of U.S.-China competition, following DeepSeek's 2025 release of cost-competitive open-source models. The brief examines definitional debates over what qualifies as truly 'open-source' AI (highlighting the distinction between open-weight and fully open models), India's Delhi Declaration advocating for open-source approaches as tools for sovereignty and inclusive development, and emerging governance questions around legal exemptions for open-source models under frameworks like the EU AI Act. The analysis positions open-source AI as a strategic lever for middle powers and Global South nations to reduce technology costs, preserve sovereignty, and enable domestic adaptation.
Why it matters
The open-source AI debate has shifted from a technical question to a geopolitical and governance challenge. Executives and policymakers need to understand how definitional choices (open-weight versus fully open) affect legal obligations, competitive positioning, and supply chain decisions as regulatory frameworks diverge globally.
Action needed
Review internal AI procurement and development policies to clarify stance on open-weight versus fully open models. Legal teams should monitor EU AI Act exemption criteria and assess how India's emerging frameworks may influence procurement requirements in Asian markets.