One thing that can be done here without saying, "all government research must be free" would be to explicitly put funding priority on labs that have a history of publishing in open access journals, or prioritize grants where the author commits ahead of time that they will publish in open access journals.

A big part of success in modern scientific research is ability to secure funding, so if you put your thumbs on the scales enough, it could catalyze a move to open access journals as the "prestigious journals."