오픈AI “ 미국 ‘AI 안전 표준’ 주도해야...한국 등 10개국과 협력 필요”

OpenAI, the developer of the GPT (Generative Pre-trained Transformer) language model, has called for international cooperation in developing artificial intelligence (AI) safety standards, as the technology advances at a rapid pace. The company believes that a global framework is necessary to ensure that AI remains under human control, and has proposed utilizing AI safety research institutes established in 10 countries, including South Korea, as a means to achieve this goal.

In a recent blog post, OpenAI CEO Sam Altman stated that the United States should lead efforts to develop global AI technology standards, as the country is at the forefront of AI innovation and holds a strategic advantage in critical sectors such as finance, trade, defense, and technology. Altman made the announcement ahead of a briefing at the United Nations' Security Council on AI and international security, scheduled for the following day.

The need for AI safety standards stems from concerns over "recursive self-improvement" (RSI), a process where AI systems can improve themselves autonomously. OpenAI warns that "fully autonomous recursive self-improvement is currently happening, and we should not proceed with it until it can be done safely." The company emphasizes that any uncontrolled self-improvement could render it impossible for humans to maintain oversight over AI development, thereby depriving us of effective control over AI technology.