I wouldn't quite trust an AI's opinion in wether given code is malicious or not, maybe in the future, but not quite yet.