Per the dialog initial focus proposal, this unresolved comment has been moved here for further discussion:

As such, authors should use the autofocus attribute on the descendant element of the dialog that the user will expect to immediately interact with after the dialog opens. If there is no such element, then authors should use the autofocus attribute on the dialog element itself.

Summary of exchange between Domenic and Aaron:

• This would mean conformance checkers will warn authors if a dialog element doesn't have autofocus specified somewhere in the dialog.
• agreement that this is probably the right thing to do, but worry about potential warning fatigue or requiring redundant autofocus attributes.

cc @sideshowbarker for thoughts on the matter