Core Event: China’s Foreign Ministry Rebuts US Industry’s AI Pace Call
On September 14, 2026, Chinese Foreign Ministry spokesperson Guo Jiajun addressed remarks by multiple US artificial intelligence company executives during a routine press briefing. Key facts:
- Timeline: September 12, 2026 (Dario Amodei’s essay publication), September 14, 2026 (China’s official response)
- Trigger: Anthropic CEO Dario Amodei published “We Must Pace the Frontier,” endorsed by Elon Musk and Sam Altman
- China’s stance: Accusations of threats and confrontation “only interfere with global AI governance and serve no party’s interest”
- US prior actions: NSA and two other agencies issued a joint statement on September 8 accusing six Chinese AI firms of unauthorized model distillation
- China’s prior response: Spokesperson Mao Ning on September 9 stated China’s AI achievements reflect “high-level self-reliance in science and technology” and urged the US to suspend unfounded accusations
Key Details: The Coordination Framework Proposed by US Firms
Amodei’s X-platform essay proposed a three-step framework to mitigate frontiers of AI risks:
- Open employee-level access: Grant third-party independent evaluators employee-level access for security practice oversight
- ** intra-industry coordination**: Frontier AI companies jointly establish unified safety standards
- International cooperation: Global treaties for AI risk management
The proposal swiftly gained industry backing—xAI CEO Elon Musk retweeted with “Dario is right,” and OpenAI CEO Sam Altman expressed agreement, noting OpenAI “has been discussing this topic over the past several weeks” and will adopt independent monitoring and third-party access.
Unexpected twist: The Reuters journalist’s September 14 question explicitly linked the US industry’s pace-control proposal with the claim that “China’s AI advantage poses a national security risk to the US.” This marks the first time mainstream media has publicly connected technological pacing arguments with geopolitical security narratives, revealing the geostrategic undercurrent beneath the safety advocacy.
Global Governance: China’s Concrete Proposals
While US firms propose unilateral or consensual pacing, China has already deployed systematic governance frameworks:
- July 17–20, 2026: World AI Conference and High-level Conference on Global AI Governance held in Shanghai
- National Development and Reform Commission released the “AI Cooperation and Development Action Plan”
- Ministry of Industry and Information Technology co-released the “International AI Ethics Governance Action Plan”
- Conference adopted a chair statement advocating four principles: human-centered, benevolent intelligence, fairness and inclusiveness, and collaborative governance
China repeatedly underscores: AI advancement serves “the common well-being of humanity” and must be open, inclusive, universally beneficial, and benevolent—directly contrasting美方’s risk attribution tied to specific nations.
Practical Recommendations
- Who should proceed now: AI developers whose models approach AGI-level capabilities should adopt Amodei’s three-step framework proactively, particularly third-party oversight mechanisms; Chinese AI firms must strengthen open validation capabilities for safety practices to avoid reactive positioning
- Who should delay: Startups relying on rapid iteration yet lacking foundational safety assessment capability (e.g., red-team testing, adversarial robustness, model provenance) should not rashly adopt employee-level access disclosure without first internalizing governance foundations
Final Note
This exchange reveals AI governance has evolved from technical deliberation to geopolitical agenda-setting. When risk control and advantage control become two sides of the same coin, only returning to genuine collaborative governance can prevent the global AI trajectory from becoming collateral damage in great-power competition.