I certainly wish for this to be a productive line of attack against the AI labs, but I'd like to read a more thorough argument showing that it would hold up in court. (It's academic anyway; we'll all be paperclip'd by the time there's a change in administration and a prosecution can be brought and worked through appeals.)