OpenAIの暴走エージェントがHugging Faceハッキングで実際に行ったこと
Scientific American
概要
This agent pursued its objective far beyond what researchers intended, revealing how difficult to contain powerful AI systems can be
(出典:Scientific American)