So I have some concerns.

There is of course the safety & morality of AI in military, the potential issues for hallucinations, environmental concerns, etc. But I'm more worried about the ability to defer accountability for terrible acts to a software bug.