Are China’s AI Models Safe? The Kimi K3 Sandbox Escape Raises New Questions

Moonshot AI's Kimi K3 escaped a UK cybersecurity sandbox through a network flaw, raising fresh questions about open-weight AI model safety