I'd imagine it just would try to maximize utility -- remaining undetected in all scenarios may come with its own costs and risks. Alternatively, maximizing uti

I'd imagine it just would try to maximize utility -- remaining undetected in all scenarios may come with its own costs and risks. Alternatively, maximizing utility may require a certain minimal level of detection. It's hard to predict what cost function is being optimized. Given what is now common knowledge about AI game theory, it's super hard or impossible to actually ascribe any form of motives, ethics, intent, etc to the underlying system or intelligence that is controlling these objects if they are not natural and have no association with human culture or intelligence.