If you have a fingerprint or can hack it, it basically will let you into the phone and into the auth app.
- It should be known only by you and
- it should be easy to change in case it is compromised.
Imagine a human guard tasked with authenticating people with a database of fingerprints. Someone comes up says that they're John Jameson, the guard takes their hand and makes an impression with some ink and paper and then compares it to impression they have on file for Mr. Jameson. Could you hack this system by lifting John's prints? Nope!
Why? Because the real trick to biometric auth isn't that your fingerprint is secret, it's that it's somewhere between difficult to impossible (depending on the metric) to produce a living breathing human with the same biometric readings as you. The strength of a biometric auth system is determined by how well it can determine that the reading it's taking is coming from a human. You don't ever have to rotate your fingerprint or your face because it doesn't need to be secret for these systems to work.
Well, I don't know the exact specifics, but that's how I would implement security for biometrics input devices. Apart from that, I seem to recall that Apple does something with validating various sensors and rejecting "unknown" sensors (although that might be based on sensor make/model/SN and not necessarily cryptographic)
> Also this will not prevent a physical identical fingerprint model attack
Correct, but no fingerprint reader that is only a fingerprint reader will prevent such an attack: No lock is made to be unopenable by (a copy of) the key to that lock. If your fingerprint is the key, a good enough physical copy of the fingerprint will open the locked device.