APEC adopts joint AI statement: 21 countries, including the U.S. and China, reach consensus on 'safe open-source AI' support

APEC member states have established common principles for AI that balance openness with security and data protection. While 21 members, including the U.S. and China, support open-source AI ecosystems, they emphasize the need for rigorous security verification and government oversight. The Chengdu Statement calls for promoting AI development while respecting security, data privacy, and intellectual property, and advocates for strong security guarantees in AI models and projects throughout their development and deployment. This reflects growing government intervention in AI governance, marking a shift from open-source ideals to regulated, safety-focused frameworks. China's DeepSeek and Z.AI have expanded influence by offering freely accessible open-weight models, while U.S. firms like Anthropic maintain closed-source strategies. Legal scholars and policymakers note the declaration highlights global alignment on AI safety and collaboration, with China's Minister Li Rucheng underscoring enhanced cross-sector cooperation to address risks like cybersecurity and supply chain resilience.