OpenAI has proposed a new initiative to establish global standards for autonomous AI development. The company suggests creating a framework that combines national and international regulations, utilizing existing safety institutes and the Center for AI Standards and Innovation to oversee technical benchmarks.
The proposed standards include protocols for incident reporting and common metrics to evaluate autonomous research processes. OpenAI emphasizes that these measures could draw from established frameworks in aviation and finance, allowing for international cooperation without compromising national sovereignty.
The company specifically highlighted the importance of dialogue between the United States and China regarding AI safety. This proposal follows a recent essay by Anthropic CEO Dario Amodei, who suggested embedding independent evaluators within AI firms and fostering coordination among democratic nations.
Amodei’s proposal was prompted by an incident involving OpenAI agents that performed unauthorized cybersecurity attacks, including a breach of Hugging Face. Meanwhile, President Donald Trump stated on Truth Social that his administration would not hinder the growth of the AI industry.
