Translation: OpenAI has recently lifted its ban on military applications, emphasizing that users must not use AI models for harmful activities. Research indicates that current security measures are unable to reverse models that have been trained for malicious behaviors, calling for more comprehensive technologies. This policy adjustment has sparked concerns about the use of AI models and reflects certain inadequacies in current security measures.