This doesn't really make sense to me, what do you mean? Client-side you do different computation for sealed sender delivery/receipt. What's your normal standard of trust that a hosted, open source project is running the same code that they've made public?
I think if they store any metadata that we don't know about, the lie runs very very deep, like to conspiracy theory levels that don't really make sense for a registered nonprofit: https://signal.org/bigbrother/