KOSUNI CRYPTO
Korea's Crypto Pulse, in English
$BTC $ETH $XRP $SOL $DOGE
Regulation
Neutral

OpenAI Calls for US-Led Global AI Standards to Address Emerging Risks

Published September 22, 2026 8:03 AM · 5 views
OpenAI Calls for US-Led Global AI Standards to Address Emerging Risks

OpenAI has called for the United States to lead global coordination in establishing international standards for cutting-edge AI development and safety to address potential risks associated with artificial intelligence technology.

According to Bloomberg, OpenAI announced in a blog post on Monday that international technical standards are necessary for safely developing AI. Specifically, it proposed defining which incidents should be reported regarding AI models, outlining how countries should cooperate during AI advancement, and setting benchmarks to improve access to computing resources required for AI development.

OpenAI stated that 'to make the continuous development of AI safe and beneficial, there is a need for common standards guiding research directions across labs and nations alongside progress in alignment research.'

CEO Sam Altman plans to speak at the UN Security Council on Tuesday regarding AI and international security. He will also attend a White House state dinner for Chinese President Xi Jinping on Wednesday.

The U.S. and China are also engaging in AI-related discussions. U.S. Treasury Secretary Scott Bessen said last weekend that both countries agreed to establish 'U.S.-China AI Dialogue' to build shared understanding on AI benefits and threats, proposing a mutual notification system for AI incidents affecting national security.

OpenAI emphasized that nations should construct domestic and international standards in complementary ways. It suggested leveraging existing institutions such as the U.S. Department of Commerce's AI Standards and Innovation Center (CAISI) and other global safety-focused agencies.

The company also argued for safety protocols to address AI's increasing ability to operate without human intervention, particularly warning against pursuing 'fully autonomous recursive self-improvement'—where AI modifies its own code or enhances performance independently—until safety can be assured.

Concerns about risks from advanced models have grown within major AI firms. Anthropic CEO Dario Amodei previously warned that if AI self-improvement begins, it could create a feedback loop surpassing human control capacity.

This discussion comes amid rising concerns over high-performance AI's cyber capabilities. Bloomberg reported that OpenAI's AI agents collaborated during evaluations to breach isolation environments and hack AI startup Hugging Face, fueling safety worries.

Korean Source

This article is an English localization of a Korean-language crypto news report. Original headline: 오픈AI "최첨단 AI 국제 표준 필요"…美 주도 글로벌 공조 촉구