I think an important point is being missed.
As others have noted, in practice it may be difficult to "prove" a theory is falsified. There is often wiggle room which may be very legitamate or just someone clinging to a pet theory for too long.
However, the point about falsification is that a theory MUST be falsifiable
in principle.
That is, the theory must be robust enough to allow for predictions based upon it that can be tested. It must have some way of tying it to some potential observations that would allow it to be giving some confirmation or disconfirmation.
A current example is the concern that string theory has no such potential falsification with current technology. It is derided by some because it can't be "tested". In this context "tested" is used exactly as "prove" is used in "The exception proves the rule." It is the potential for a "test" which can, of course, be failed that is why falsification is a central and esential part of science.