At the moment, any user-provided string inputs (i.e. passwords, additional data, etc.) is used as is, without any form of Unicode normalization. In order to ensure security and portability, all input strings (that are a superset of ASCII) should be normalized according to a canonical rule. See the following two issues which stemmed this issue: * <https://github.com/FiloSottile/age/discussions/601> * <https://github.com/samuel-lucas6/Kryptor/discussions/78>