Because RU and AU are extensionally equivalent: To say that two ethical theories are "extensionally equivalent" is to say that they imply exactly the same judgments about every possible case, so that something will be right according to AU if and only if it is also right according to RU. Why think this is true? For any set of rules you are considering, no matter how complex the rules already are, if you can think of a case in which violating the rules would produce more utility than obeying them, then you have a utilitarian reason for amending the rules to allow that exception. And that goes for any possible case, including, e.g., punishing the innocent. So, it is claimed, the RUian ought to choose a set of rules which implies exactly the same judgments as an AUian would reach.
Because RU "collapses into" AU: Others have argued that the RUian should choose act utilitarianism as the rule to follow. After all, "So act as to maximize aggregate happiness in every situation" is one possible set of rules, and if people followed that rule forever, they presumably would produce the maximum possible happiness.
