The short version
- Chinese officials and analysts interpret US calls for a global slowdown in AI development as strategic moves to preserve American market leadership rather than genuine safety concerns.
- Beijing has released an updated governance framework that addresses specific technical risks like autonomous agents and recursive self-improvement, signaling a shift toward controlling AI actions rather than just outputs.
- While both nations acknowledge the dangers of advanced systems, structural differences in regulation and state control create divergent approaches to balancing innovation speed with safety protocols.
Tensions between Washington and Beijing over artificial intelligence governance have intensified following recent warnings from American tech leaders about the potential catastrophes associated with losing control of frontier models. Anthropic chief executive Dario Amodei recently advocated for a worldwide pacing mechanism for cutting-edge AI development, explicitly stating that such cooperation would require engagement with China. However, Chinese officials and industry observers have pushed back against these proposals, viewing them not as collaborative safety measures but as strategic efforts to lock in United States technological advantages.
Analysts note that while American models currently hold a lead in frontier capabilities, the gap is narrowing. Beijing considers AI a core component of national sovereignty and is unwilling to jeopardize its progress by adhering to slowdown terms dictated by US firms. Lian Jye Su of the technology research group Omdia explained that China perceives these calls as bids to maintain American dominance. Amodei’s comments reinforced this perception, as he argued for maintaining democracies’ AI lead over autocracies and supported continued export controls on advanced chips and machinery.
In response to Amodei’s essay, China’s foreign ministry criticized the approach, stating that fearmongering and confrontation would only disrupt global AI governance. Despite this diplomatic friction, experts suggest there is more common ground than publicly acknowledged regarding the underlying risks. Gabriel Wagner, a researcher at Concordia, a Beijing-based AI safety consultancy, noted that both sides are grappling with similar concerns about system stability and control, even if their political rhetoric differs significantly.
China recently released the third iteration of its AI safety governance framework, which serves as a detailed mapping of the risk landscape. This update includes specific provisions for risks posed by autonomous agents, cybersecurity threats from frontier models, and the hypothetical scenario of recursive self-improvement. Leia Wang of the Carnegie Endowment for International Peace observed that Chinese regulation is iterative, allowing the government to adapt quickly to new developments. This contrasts with the more static regulatory environment in some Western jurisdictions.
The focus of Chinese regulation appears to be shifting from controlling what AI systems say to controlling what they do. An algorithm registry established in 2022 has been supplemented by warnings regarding agents, reflecting a broader concern about actionable outputs rather than just textual generation. Additionally, national regulations issued this year aim to curb emotional dependency on companion AIs, highlighting the government’s attention to social stability and psychological impacts alongside technical safety.
Despite these regulatory efforts, Chinese companies engage in less voluntary monitoring and evaluation compared to their US counterparts. Wang pointed out that neither government seems fully prepared for a loss-of-control incident of the magnitude outlined by recent Silicon Valley warnings. However, China’s centralized authority over critical infrastructure and commercial activity may position it better than the United States to address potential AI disasters through direct state intervention.
The Chinese Communist Party remains highly attuned to risks that could threaten its political control, ensuring that new technologies develop within state-defined boundaries. This proactive stance is evident in the encouragement of AI adoption across the economy, which has been enshrined in the latest Five Year Plan. Local governments are offering funding and low rents to tech companies, fostering an environment where innovation is actively promoted alongside strict oversight.
Entrepreneurs like Boris But, who runs an education company, have embraced AI to remain competitive, developing tools that help clients navigate information previously sourced from rival firms or chatbots. His experience reflects a broader trend in China, where businesses are urged to integrate AI to avoid obsolescence. Factories across the manufacturing base are also being encouraged to convert their processes for the AI era, demonstrating the economic imperative driving adoption.
President Donald Trump has dismissed fears that AI could wipe out humanity as a hoax, contrasting sharply with the cautious tone emerging from Silicon Valley. With Trump set to meet Chinese President Xi Jinping in Washington next week, AI is expected to be on the agenda. However, analysts remain skeptical that a major breakthrough or agreement will emerge from these discussions, given the deep-seated strategic mistrust and differing regulatory philosophies.
The divergence in approaches highlights a fundamental challenge in global AI governance: balancing the need for rapid innovation with the imperative of safety. While the US emphasizes voluntary industry standards and export controls, China relies on state-directed regulation and centralized oversight. Both systems face uncertainties regarding how to manage increasingly autonomous systems, leaving the international community without a unified framework for addressing potential existential risks.
Sources behind this briefing
Go to the original reporting
- The Guardian World↗Why China is pushing back on US warnings over rapid AI development
