Update: turns out it's a known bug on their back end service (they're working on a fix). Everything now sorted out manually.