Trying to make sense of Flatpak's threat model is just near-impossible. It might protect you from something, if the specific app's configuration told it to do so. Starting a random Flatpak app, you have no guarantees, and the UX doesn't communicate anything about this.