OpenAI has called for the United States to lead an international effort to develop technical standards for frontier artificial intelligence, including standards for recursive self-improvement (RSI).
The company said common standards are needed as AI systems become more capable and autonomous, particularly as they take on more work in AI research and development.
“People outside the AI labs should have a real say in how this technology develops, and a clear way to judge if it’s happening safely,” OpenAI CEO Sam Altman said in a post on X.
People outside the AI labs should have a real say in how this technology develops, and a clear way to judge if it’s happening safely.
Standards should help prevent the concentration of power, including by making sure new companies and open-model companies can compete.
They…
— Sam Altman (@sama) September 22, 2026
He said standards should also help prevent the concentration of power, allow new companies and open-model companies to compete, and help countries and companies compare evidence and learn from failures.
OpenAI said the US should lead the effort because its AI industry is at the technical frontier and the country has a central role in global finance, trade, defense, technology and information systems.
The company said fully autonomous RSI is not happening today and should not be pursued until it can be done safely while preserving human control. It warned that without adequate safeguards, increasingly autonomous AI research could make it harder for humans to maintain practical control over the development process.
OpenAI said the Hugging Face incident, while not a direct result of RSI, offers a preview of the kinds of risks that could become more severe without robust safeguards and alignment.
The proposal comes as the UN Security Council prepares to discuss AI and international security, with Altman and other AI industry leaders scheduled to brief the council today.
OpenAI said automated AI research could allow AI systems to contribute to the development of successive generations of AI. It also said this could lower the cost of advanced intelligence and support AI safety research.
The company identified three challenges requiring international cooperation: fragmentation, collective-action problems and uneven technical capacity across countries. It proposed using the emerging network of AI safety institutes in Australia, Canada, Germany, France, Kenya, Japan, South Korea, Singapore, India and the UK to help develop common standards.
The proposed framework would establish common methods for measuring AI capabilities, evaluating risks and determining whether safeguards are sufficient. It would also cover human oversight of automated AI research and the classification, tracking and reporting of AI-related incidents.
OpenAI said the standards would apply to both open and closed AI models. It stressed that they would not amount to licences, mandatory pre-release reviews or approval requirements, with individual governments deciding whether and how to incorporate the standards into national laws.
The company said international standards could help address differences between national AI rules while giving stakeholders outside AI labs a greater role in shaping how the technology develops.
OpenAI also called for dialogue between the US and China on AI safety, saying such engagement would be a positive step and that upcoming talks come at an opportune time.
The proposal comes as Washington and Beijing discuss an AI dialogue ahead of Chinese President Xi Jinping’s planned September 23-25 visit to the US, with US Treasury Secretary Scott Bessent saying Washington had proposed a notification mechanism for AI-related incidents that rise to the national-security level for Trump and Xi to consider at their summit.
OpenAI said “pacing” AI development should not mean maintaining a fixed speed of progress. Instead, it argued that alignment research and deployment should stay ahead of AI capabilities.
Also Read: China AI Boom: How is China Financing it? What a US-Based Research Firm Says
