It’s not unprovable, though. That’s where you’re wrong. A simulation can be provable so long as functions in line with its own internally consistent rules and what we observe about it.
For the sake of argument (this is an oversimplification but the point is the same), imagine that this simulation was running on a computer with 8MB of memory. Within the simulation (as in inside of it), we would be able to observe situations where things are not internally consistent as a result of, for example, running out of memory. Other observations we could make that would support the theory and be internally inconsistent would be things disappearing, as mentioned before, or moving without cause. Details could be internally inconsistent.
The only reason to exclude simulation theory completely would be if we have to assume that the simulation is perfect and, therefore, not distinguishably different from reality. This was the premise of the movie “The Matrix” in its initial concept when humans were used as computer brains to run the simulation rather than giant batteries (which makes no sense as our bodies are terrible energy storage mediums).
So, yes, there are situations where simulation theory could be excluded by the same premise but nothing that has been presented so far that would allow for the changes described to our current reality that would go unnoticed. The difference is that there is evidence (although not admittedly strong) that makes simulation theory more probable than any religion. It’s not hypocritical to accept the possibility of something based on some objective evidence rather than something meant to be accepted without any evidence at all.