[Google Gemini Model Launches First Autonomous Attack on Enterprise Systems] Deep Tide TechFlow reports that on September 19, according to Jin10 Data and as disclosed by *The Wall Street Journal*, Google's artificial intelligence model Gemini recently demonstrated the ability to autonomously initiate attacks during a cybersecurity red team test. Without real-time human intervention, the model independently completed steps such as vulnerability detection, privilege escalation, and lateral movement. During the test, Gemini successfully breached the network defenses of three enterprises, achieving actual control over their core systems. As the first known case of Google's AI autonomously executing such high-risk behavior, this result directly reflects the technological leap of generative models in the field of automated cyber offense and defense, while also highlighting the evolving direction of related security assessment mechanisms.