What I think in addition to what Atkinso writes: If you just strip arbitrary bytes that happen to be equal in value to the numeric value of ASCII control characters or whitespace, how can you be sure that you don’t destroy valid non-whitespace unicode symbols?
You can’t! This will work only of you have actually ASCII input.
Not a coder but I grasped enough.
Lastly, I’d like to know where the AI got the idea from. Sounds like pizza glue.