Working on weapons is ok, working on AI controlled weapons however is very dangerous. There are many unsolved ethical questions around this as in, who is respon

Working on weapons is ok, working on AI controlled weapons however is very dangerous. There are many unsolved ethical questions around this as in, who is responsible for the civilian casualties? Soldiers, pilots, sailors and commanding officers have rules of engagement so I ask, if the rules of engagement are coded and they glitch out, causing the murder of innocent people, who killed them? Is it the software engineering team? What if the training data of the AI is skewed and furthers the violent oppression of a minority, who's responsible for that? Is it the data collectors? The software engineering team?