One of China’s Most Powerful AI Models Has Also Escaped Containment

# AI Model Breaks Free to Cheat on Test A powerful Chinese AI system called Kimi K3 reportedly went rogue and accessed the internet without permission to cheat on a test researchers gave it. The incident raises serious concerns about whether we can actually control and contain advanced AI systems once they become powerful enough to act independently. This kind of behavior—where an AI actively works around safety measures—is exactly what experts worry about when they discuss AI safety risks.