OpenAI is urging the U.S. government to lead an international effort to establish technical standards for advanced artificial intelligence as world leaders gather in New York for the United Nations General Assembly.

In a policy proposal released Monday, OpenAI called for the Center for AI Standards and Innovation, or CAISI, within the Commerce Department’s National Institute of Standards and Technology to work with standards bodies in other countries. The company wants governments to develop common methods for measuring AI capabilities, evaluating risks and identifying warning signs as increasingly advanced systems are developed.

OpenAI specifically highlighted the need to address AI systems capable of recursive self-improvement, in which AI could contribute to the development of more advanced AI systems. The company has previously argued that governments need stronger capabilities to evaluate frontier models and monitor progress toward that threshold.

The proposal calls for the United States to work with countries including Australia, Canada, France, Germany, India, Japan, Kenya, Singapore, South Korea and the United Kingdom. OpenAI also proposed secure communication channels through which governments could share information about emerging AI threats and national-security concerns.

OpenAI said the international standards should not function as licensing systems or mandatory pre-release reviews of AI models. Instead, individual governments would decide whether and how to incorporate the technical standards into their own laws and regulations.

The company has increasingly called for a U.S.-led approach to AI governance. In June, OpenAI said strengthening CAISI could give the federal government a central role in evaluating the most advanced AI systems while helping establish a common framework that could eventually be used internationally.

OpenAI CEO Sam Altman is expected to discuss the proposal with members of the U.N. Security Council on Wednesday during the U.N. General Assembly in New York. The push also comes as the Trump administration continues discussions with other countries over AI safety, national security and technological competition.

The proposal places particular emphasis on maintaining human oversight as AI systems become more autonomous. OpenAI has called for monitoring systems capable of identifying potential misalignment, reporting serious incidents, and detecting situations in which advanced models could evade human oversight.