OpenAI calls for US-led global frontier AI standards

OpenAI is calling for the United States to lead an international effort to create technical standards for frontier AI, including shared rules for evaluating models, reporting incidents and managing recursive self-improvement.

The company frames the proposal around automated AI research: systems that increasingly help develop later AI systems. OpenAI says fully autonomous recursive self-improvement is not happening today and should not be pursued unless it can be done safely, with human control and democratic choices about the risks.

The standards push would use national AI safety institutes and the Center for AI Standards and Innovation as part of a broader network for measuring frontier capabilities, assessing safeguards and defining when automated research processes need immediate human review. OpenAI says this work should include both open and closed model developers, independent experts and academia.

One important caveat: OpenAI says these standards should not be licenses, mandatory prerelease review or model approval requirements. Instead, national governments would decide whether and how to incorporate common technical standards into their own legal systems.

The broader message is political as much as technical. OpenAI argues that without common measurements and incident reporting, countries could end up with conflicting AI safety rules and weaker responses to cross-border risks. Reuters summarized the proposal as a call for Washington to lead a global standards effort for frontier AI.