Most actions that are on the record, with classified data properly conveyed through their "high side" inboxes and properly archived, will have those records accessible to special counsel or historical analysts. If, as I suspect, most of the current cabinet's principals committees are meeting over Signal, the records of those communications will be conspicuously absent.
It's not like these guys are masterminds meticulously generating compelling and consistent alternate records in the SCIF, then also pulling out their phones and telling most of the same people most of the same things in group chat messages. They're just not having the discussions in the SCIF at all, and that will be evident to anyone who cares to investigate.
I don't know the latest details about Android/iOS app signing, but presumably reproducible builds + sufficiently strong signing would make it secure enough for most users. For those who are truly paranoid, then can build it themselves (subject to their own device OS's requirements, which are hardly a unique problem to Signal).
In short, Signal's security should be as good as any mobile app can be, and can be even better if you're willing to put in legwork.
When was the last time you checked what updates have been made to the git repo?
Of course what you're saying is "technically" possible to avoid signal changing code and circumventing encryption, but show me one person who does a check of all the changes to signal source(and verified the binary matches) before they let their app store update it and they launch it...
Everyone I know has signal auto-update through the app store and don't even know it updated until after they launch it.
They haven't, but if they decide they want to what's technologically stopping signal from:
1. Making an update that doesn't exist in git which pushes decrypted messages to their server when you launch the app
2. Push this update to the app store
If a government ever compelled them to?
I understand the point about the app being distributed by the same people who run the service, but it's much harder to hide shenanigans with a local app versus a web app, especially when the app is open source.
All I'm getting at is that any company that distributes code to you and tells you they can't see your data is lying. They just don't want to access your data right now.
I would suggest people understand this and position themselves accordingly security-wise.
If that means not using signal because its not secure enough then ok.
If that means continuing to use signal with the understanding that it's only secure until signal decides they want your data(or a gov forces them to), then ok
Splitting management of an app and service is the exact solution. If signal can't control when to push updates to your phone then they can't control when they want to break encryption.
In your compromised browser example we understand that browsers have an interest in imementing HTTPS correctly and treat them accordingly. That's part of the reason the market is dominated by 2 engines that do their development as much in the public as possible
They could at any point push an update that decrypts your messages locally and pushes them decrypted to a server. The only way to prevent this would be to verify each binary update to signal matches the source code, and no modifications have been made to the source to do this.
Is that part of your "signal update" routine?