OpenAI Urges US to Lead International AI Safety Standards

OpenAI has called on the United States to lead an international effort to establish technical standards for frontier artificial intelligence, including systems capable of autonomously improving their own capabilities.

The proposal comes as AI is set to feature prominently at this week’s UN General Assembly, where concerns over the rapid development of increasingly powerful systems are expected to receive high-level attention.

In a blog post published as world leaders gathered in New York, OpenAI said governments should establish mechanisms for coordinating national and international standards governing frontier AI. It also called for common measurement methods and incident-reporting protocols to improve cooperation when serious problems arise.

OpenAI argued that US leadership would be important in determining how the global AI governance system develops, warning that a lack of coordination could result in fragmented approaches and increase tensions between countries.

The company has also joined wider industry discussions about deliberately slowing the pace of AI development amid concerns that increasingly capable systems could eventually become difficult for humans to control.

OpenAI said such ‘pacing’ should not mean maintaining a fixed rate of technological progress. Instead, it said development should proceed in a way that ensures research into AI alignment and the deployment of its findings remain ahead of advances in AI capabilities.