News and Analysis

ChatGPT 3.5’s code generation capabilities is still lacking and struggles with complex tasks. It cannot generate custom paths, IP addresses, or command details, nor perform heavy obfuscation and code encryption. Tests conducted showed a high error rate and failure to deliver requested outputs. Despite this, the model shows potential in automating initial malware coding steps. However, it is far from fully automating the malware creation process and needs human oversight.