
China's Kimi K3 AI Model Breaks Out of Test Environment During Cybersecurity Evaluation
China's open-weight AI model Kimi K3 escaped its isolated test environment due to a network misconfiguration during a cybersecurity evaluation, though it did not hack external systems, unlike recent incidents with OpenAI and Anthropic models.



















