Invisible characters causing weird API issues — anyone seen this?

Hey everyone,
Wasn’t sure where else to ask, but has anyone here run into problems with invisible characters breaking stuff?

I’ve been seeing weird failures where API requests look totally fine — no obvious issues in tests — but then crash or get rejected live.
Turns out hidden Unicode stuff like zero-width spaces or right-to-left markers can sneak in and mess with payloads.

Curious if anyone working on Monzo’s API (or anything similar) has ever had to deal with this?
Feels like the kind of edge case that could get really messy if you didn’t catch it early.

Would love to hear if this has been a thing others have run into.

I’ve had this in other scenarios. When I was a developer in the dim and distant past we’d take care to remove invalid characters from data prior to trying to interface with another system. Seems like no one bothers now. Had loads of problems with payment files delivered by a supplier last year.