Someone who programmed the AI committed homicide using/programming the AI. But then... what about bugs? What if these bugs are not incidental? Let us assume that they are. Would it count then as non-premeditated murder? Who are the people who reviewed the code in this case, then? Accessories?