"Every" is a bit rich, considering we're likely 100s of years from testing it directly.
Doesn't that also mean that (in such a hypothetical situation) our best theory says that our previously best theories will give accurate predictions up until the point at which we observe something equivalent to a galactic-scale collider? If so, that doesn't sound so bad to me.
On the bright side, if there was a single particle that didn't move like a string, string theory would be falsified. On the dim side, it's difficult for anything to not move like a string, because string theory is so powerful as a framework that you can get it to make just about anything consistent. What you really want in physics is a brittle framework that only allows you to slot in a few different behaviors.
It's kind of funny because programmers want the opposite thing that physicists do. When your client asks for a behavior that you have painted yourself into a corner against ever implementing your software has been "falsified," an expensive disaster that you work very hard in advance to prevent! If you made universes for a living, strings would be an ideal framework.
For example, (going from memory here, so don't quote me), your basic (Von Neumann's?) formulation of quantum mechanics was basically a generalization of Newtonian physics that specified an unreasonable number of non-real possibilities, and you can only recover meaningful results by taking Newtonian physics as an axiom a la the 'classical limit'. This would have just been mathematical over-complication had it not been for the fact that this model predicted things that differed from Newton physics in measurable ways.
So to reiterate, if your theory is unfalsifiable, it is not impressive that it agrees with every measurement. All you'd have achieved is restating what is already known in a language nobody wants to use.