Hi @CCP_Aurora we will have a look, I recall getting the regexes to work properly in unicode and handle boundaries correctly was a bit of an adventure. This certainly looks like a bug.
@gerhard may have some ideas as well, I recall he worked on similar issues in the past.