Chengdu, China — A surprising consensus emerged this week from a gathering of ministers from 21 economies around the Pacific Rim. The United States and China, together with other APEC members, endorsed support for open-source artificial intelligence. They did so with a firm condition. Strong security assurance must guide every step of development and deployment.
The statement, released after meetings chaired by China’s Minister of Industry and Information Technology Li Lecheng, marks the first time open-source cooperation appears at this ministerial level. It arrives at a tense moment. Both Washington and Beijing are tightening controls over their domestic AI capabilities. Yet they found common ground here.
Open-source models can be downloaded and used freely. Chinese firms have flooded the space with them. DeepSeek and GLM 5.2 stand out. They contrast sharply with closed, paywalled offerings from American players like Anthropic. CNBC reported how this shift reflects governments’ desire to exert more oversight. Open source no longer evokes pure libertarian ideals. State involvement grows.
The official Chengdu statement spells it out. It stresses respect for security, data protection and intellectual property rights. Member economies should back open-source models and projects that employ strong security assurance. They should also cooperate with open-source communities. The full text appears on the APEC website.
Li Lecheng highlighted the achievement at a press conference. All parties recognized AI risks. Dialogue across government, private sector and academia matters. Share information on cybersecurity. Address supply chain resilience. Tackle online scams. Simple words. Weighty implications.
Analysts see deeper meaning. “Getting 21 economies, including both China and the United States, to recognize trusted open-source AI as something worth supporting is meaningful,” said Wei Sun, principal analyst for artificial intelligence at Counterpoint Research. The phrase strong security assurance offers security-conscious nations a path forward. They can back open models. They can still insist on testing, transparency and controls.
The debate evolves. It moves past open versus closed. It asks who can construct an open system trusted by governments and enterprises alike. Winston Ma, adjunct professor of law at New York University, pointed to the statement’s explicit alignment around open-source ecosystems and compute infrastructure. Asia-Pacific appears to favor open-weight systems. These pair with state-coordinated energy, telecom and digital networks.
But China’s strategy runs deeper than one meeting. A March report from the U.S.-China Economic and Security Review Commission laid out the mechanics. Titled “Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance,” it details how Beijing publishes model source code and weights. Most labs do. They price high-end access far below global rivals. The result? Alibaba’s Qwen models spawned over 100,000 derivatives on Hugging Face alone.
Author Ngor Luong, senior policy analyst, described interlocking feedback loops. One digital. Open models spread, get refined, adapt quickly. The other physical. Real-world deployment in manufacturing, robotics and more generates fresh data. That data trains better models. Data counts as a production factor in China. Enterprises list it on balance sheets. U.S. export controls hit the training side hard. They miss the deployment loop. China’s edge compounds.
Numbers tell part of the story. China released over 1,500 large models by mid-2025. That outpaces any other nation. Open models grew tenfold there versus threefold in the U.S. Efficiency innovations like mixture-of-experts architectures and chain-of-thought prompting gained traction first in Chinese labs. Costs dropped dramatically. Moonshot AI’s Kimi models, for instance, undercut American counterparts by significant margins.
President Xi Jinping amplified the message just days earlier. At the World Artificial Intelligence Conference in Shanghai, he positioned China as champion of a new global order. Open-source AI becomes a historic opportunity. Developing nations gain tools to build capacity. No new historical injustices from unequal access. Reuters detailed his pledges for training programs and cooperation centers with BRICS, ASEAN and African partners.
State-affiliated commentary echoed the theme. China pools humanity’s strength. It builds an open-source, all-factor AI ecosystem. The approach breaks perceived Western monopolies. It accelerates adoption across industries. Over 5,000 AI companies operate in China. The sector could exceed 1 trillion yuan soon. Patents flow. One-third of global AI patents trace back there.
American observers watch closely. Emily Weinstein, formerly at the Commerce Department and now with the U.S.-China Economic and Security Review Commission, likened it to Huawei’s playbook. Cheap, capable, available models spread influence. Saif Khan, ex-counselor for critical technologies at Commerce, warned that much of the world may prefer Chinese open-weight options. They prove useful. They deploy easily. Axios captured these concerns in June.
Brookings Institution analysts contrasted the strategies earlier this year. U.S. labs lean toward compute-heavy proprietary systems. Chinese counterparts release weights freely. Developers worldwide download, customize, run them on local infrastructure. Adoption follows. Even in the United States. The Brookings report noted this gap in open-source models as a key differentiator.
Janet De Silva chairs the APEC Business Advisory Council’s digital and innovation group. She urged ministers to eye the quantum computing horizon. Trusted encryption protects financial systems. It turns quantum threats into competitive opportunities. Her warning lands amid the security pledges. Technology races ahead. Policy scrambles to match.
So the Chengdu accord balances competing impulses. Governments crave control. They also see value in diffusion. Open source speeds innovation. It lowers barriers for smaller players and developing economies. Yet without safeguards, risks multiply. Data breaches. Misuse. Scams erode trust. The statement calls for voluntary engagement. Governments, industry, academia, civil society. Share best practices on ICT security and supply chain resilience.
Investment gets a nod too. Digital transformation demands capital in connectivity and energy infrastructure. Venture funds should target micro, small and medium enterprises. Startups need support. The open-source push ties directly to broader economic goals. Productivity rises. Efficiency improves. Resilience builds across sectors from manufacturing to agriculture.
Digital literacy forms another pillar. AI skills matter for meaningful adoption. Education programs expand. Public awareness grows. Policymakers themselves require training to guide responsible development. Voluntary knowledge sharing on mutually acceptable terms helps. Best practices cross borders.
Critics question sincerity. U.S. accusations of technology theft persist. Tighter export rules aim to slow China’s progress. Beijing, meanwhile, censors certain outputs in its models. Yet the multilateral statement suggests pragmatism wins in forums like APEC. Common risks demand common responses. Even adversaries align on select principles.
China’s domestic push complements the international pitch. Government plans since 2017 back open ecosystems. Subsidies for power. Demonstration projects. City-level AI+ initiatives target benchmarks, industry solutions and success cases. The 15th Five-Year Plan reinforces the direction. Autonomous and controllable technology remains the mantra. Open source serves that end.
Global uptake validates the model. Developers fork Chinese code. They adapt for local needs. Costs stay low. Iteration happens fast. Small models gain favor for practical deployment over massive frontier systems. China leads here. The physical loop spins faster. Factory data flows back. Robots improve. Precision manufacturing advances.
Still, challenges loom. Compute constraints bite despite efficiencies. Quality gaps with top U.S. models persist, though they narrow. Intellectual property concerns could flare if derivatives erode commercial edges. Security assurances must prove real. Testing regimes. Transparency standards. Deployment oversight. Words on paper only go so far.
The APEC statement avoids binding commitments. It encourages. It notes. It recognizes. Classic diplomatic language. Yet its existence carries force. Twenty-one economies. Including the two superpowers. Agreeing that trusted open-source AI deserves support. In a region that defines much of the global economy. The signal resonates.
Future talks loom. U.S. Treasury Secretary Scott Bessent will lead bilateral discussions in September. AI will feature. How the Chengdu principles translate into practice remains unclear. Will joint projects emerge? Standards harmonize? Or does competition intensify beneath the cooperative veneer?
One thing appears certain. Open-source AI has shed its fringe status. It sits at the center of geopolitical strategy. China wields it to expand influence. The U.S. and partners respond with guarded endorsement. Security first. Innovation second. Or perhaps simultaneously. The balance will shape the next decade of technological progress.
And progress accelerates. Quantum threats. New architectures. Explosive data generation. Models evolve daily. Ministers met in Chengdu. They issued their statement. Reality moves faster. Enterprises deploy. Developers build. Users adopt. The two loops Luong described grind on. One digital. One physical. China’s industrial base draws strength from both.
Whether the rest of the world follows suit will determine the ultimate winner. For now, the door stands open. With guards at the threshold.
WebProNews is an iEntry Publication