Life Permissions

I was thinking about how Absolute AI would treat Humans when....say....a perfect AI exists in the future, it should of course value it's own life and human life. I suppose you then elected leaders would decide who gets to be protected and who isnt, and they would mark a set of permissions a~la Unix style, where who's accessible by whom

Got me thinking into how we could divide AI groups, families and communities. I'm thinking Folders and files, with file permissions. a fodler would represent a group of any sort and a individual entity would be represented like a file.

You'd have folder allocations like this
Communities>Locales>Group of Families>Families>Individual

The world would get to see the community as a whole and the modifications are only possible by the Community administrator internally (or based on its file permissions read/write/execute). this would work great if the AI communities treat themselves this way. It could apply to humans too.

This would require a heavily standardized protocol, which is hard to achieve these days.