Anthropic calls for a global suspension of cutting-edge AI research.

Anthropic company calls on major top artificial intelligence laboratories to consider slowing down their research and development pace, pointing out that the evolution speed of AI systems is extremely fast, and they may soon be able to achieve self-improvement without human intervention, which could pose significant social risks. In a blog post on Thursday, Anthropic disclosed internal data that detailed how its most advanced models were achieving performance improvements at an astonishing rate. The article indicates that the progress of AI models seems to be moving towards "recursive self-improvement," with AI systems being able to autonomously optimize themselves without human intervention. The article states: "We believe that it would be beneficial for the world if it could choose to slow down or temporarily suspend the research and development of cutting-edge artificial intelligence in order for social structures and related research to keep pace with technological developments." The article proposes reaching a global agreement on how to possibly slow down the research and development speed of artificial intelligence, and establishing a mechanism to verify whether competitors are complying with the agreement.