[bumby]

The memorandum calls for this to apply to unclassified research funded "wholly or in part" by federal funds. So, even a partial contribution by any of those mechanisms you identified may mark it for open access, although individual agencies are tasked with developing their own implementation policies.

What I'm afraid of occurring is over-leveraging classification of research as a means around this, if for no other reason than it's easier to implement/publish.