OpenAI Proposes US-Led Global AI Standards Ahead of China Talks

OpenAI has called on the United States government to lead the development of global AI safety standards, proposing that existing bodies coordinate international technical frameworks as American and Chinese officials prepare for high-stakes diplomatic discussions this week.

In a blog post published September 21, the company urged Washington to leverage the Center for AI Standards and Innovation (CAISI) and collaborate with other nations to establish technical standards for frontier AI systems. The proposal specifically addresses recursive self-improvement (RSI)—the hypothetical point when AI systems can autonomously develop new models without human intervention. While fully automated RSI does not yet exist, OpenAI argues that establishing standards now would create a foundation for safe development.

Why it matters

The timing reveals OpenAI's strategic calculus: positioning US-led standards as both a safety mechanism and a competitive advantage in the race against China. If Washington can broker international cooperation on AI governance, American companies may secure favorable terms for global AI adoption—potentially determining which nation's technology becomes the default choice for businesses and governments worldwide.

The Standards Framework

OpenAI's proposal centers on technical standards for capability measurement, risk assessment, and safeguard evaluation rather than mandatory licensing or pre-release approval requirements. The company envisions national governments incorporating these voluntary standards into their own legal systems as they see fit.

CAISI, housed within the Commerce Department, currently tests new AI models for the Trump administration. The agency participates in the International Network for Advanced AI Measurement, Evaluation and Science alongside organizations from nine other governments—a coalition OpenAI believes could address the fragmented understanding of AI safety across countries.

"The United States is well positioned to lead because its AI industry is at the technical frontier, and it still stands in a privileged global network position in critical areas such as finance, trade, defense, technology, and information systems," OpenAI stated in its post.

The China Question

Whether Beijing will accept US leadership on AI standards remains deeply uncertain. Chinese President Xi Jinping has advocated for international AI cooperation, but on China's terms. At the World AI Conference in Shanghai this summer, Xi announced the World Artificial Intelligence Cooperation Organization with 29 member nations including Brazil and Russia.

"AI development should not be a solo performance by a single country, but a symphony of international cooperation," Xi said at the conference.

Treasury Secretary Scott Bessent said US officials have proposed an emergency notification mechanism allowing both countries to alert each other to AI incidents posing national security threats. Xi and Trump are scheduled to meet Thursday to discuss AI safety, with OpenAI CEO Sam Altman addressing the UN Security Council on Wednesday before joining the US-China state dinner.

Political Context

OpenAI's push follows Anthropic CEO Dario Amodei's recent essay calling for government intervention to slow AI development. Altman endorsed Amodei's position, and OpenAI executives met with lawmakers to support two bills establishing federal AI safety frameworks.

The slowdown proposal drew fierce criticism from President Trump, who argued that unilateral US restraint would allow China to surge ahead. This concern makes globally coordinated measures involving Beijing essential for any regulatory framework to gain traction in Washington.

Beyond China, the Trump administration's confrontational trade policies have strained relationships with traditional allies, potentially complicating efforts to build international consensus around US-led standards.

The details were first reported by Gizmodo.