A batalha da IA Open Source já começou, e quem deve ganhar é a China:
Regulating Open Source
ANDREW POLK
Sinica, JUL 26, 2026
Xi Jinping’s speech at last week’s World AI Conference in Shanghai was an exercise in “balancing development and security.”
Xi expressed full-throated support for AI development, while indicating that regulation is moving up Beijing’s priority list.
Read together, the message was that China intends to compete hard on AI, but on its own terms, with stability ranked ahead of any near-term gains.
The bit that attracted the most attention was Xi’s reaffirmation of China’s commitment to open-source AI.
There has been growing speculation that AI safety concerns may force Chinese regulators to disallow ongoing release of open-source models.
But Xi’s message was that China isn’t going to walk away from open source.
That’s likely because the global popularity of Chinese open-source tools is a once-in-a-generation opportunity to bring the international community onto China’s AI stack – particularly in the absence of cheap, open US alternatives.
Beijing intends to press that advantage for as long as it can.
Running alongside the open-source pitch, however, was a notable elevation of regulation, ethics, and governance in Xi’s remarks.
The speech raised deep questions about practical and existential AI risk and called for new laws and risk-monitoring systems.
Beijing was already an early and fast mover on AI regulation, but Xi’s talk indicated that China will double down on AI safety, potentially placing stronger controls on both the release of new open-source models and the distribution of open model weights.
Then there was the announcement of the World Artificial Intelligence Cooperation Organization, formally launched on the eve of the conference last week.
This represents Beijing’s bid to compete with the US on global AI governance and offer the Global South an alternative to Western-led bodies.
Whether it gains traction is an open question – China’s track record on tech governance organizations is patchy at best, and the recent past is littered with abandoned Chinese-led cooperative platforms.
Zoom out from the individual announcements, though, and a bigger picture emerges: Xi believes that while AI is deeply important and transformative, it should never become so all-consuming that policymakers lose sight of what really matters.
Technology is not an end unto itself – instead, AI should serve social development and stability.
The implications of that framing are far-reaching. Beijing is prioritizing AI’s long-term development over a sprint race to build the most advanced frontier model.
Viewed through that lens, the tighter regulations, willingness to slow model releases, and the hard line on NVIDIA chip imports all begin to make a lot more sense.
The bottom line: The AI race won’t be decided this year – it is a decade-long play for who owns the AI stack that the developing world runs on, who sets the rules of global AI governance, and ultimately, whose industries make the best use of these tools for economic gain.
Kendra Schaefer, Partner and Head of Tech Policy Research, Trivium China
Nenhum comentário:
Postar um comentário