Unsupported unicode characters sent via SOAP API

I am using the SOAP API with .NET and am calling the method "findPrincipalByToken".

This works successfully when I have set up Crowd to connect to an internal Crowd Directory. However when linking to our corporate Active Directory, this call throws the following error:

{"'', hexadecimal value 0x02, is an invalid character. Line 1, position 2918."}

This is a serialization error due to the fact that unsupported unicode characters (according to the XML/SOAP specifications) are being sent via the API. This is probably due to the fact that user attributes in our Active Directory include encrypted values, such as security descriptors and signatures (i.e for MS Exchange) which we cannot modify.

It is possible for certain clients to change the default serialization behaviour (i.e to normalize these characters), however any invalid and unsupported characters should be filtered out BEFORE being transmitted via SOAP. Is this possible?

Only the following characters should be premitted (see http://www.w3.org/TR/REC-xml/)

Char ::= #x9 | #xA | #xD | x20-#xD7FF | xE000-#xFFFD | x10000-#x10FFFF