When an administrator creates a content filter policy to take an action when an email message contains a word belonging to a specific dictionary, a known issue arises depending on the type of character that follows the dictionary word:
This behavior is related to how SMG handles word boundaries. Non-ASCII characters are treated as word boundaries, effectively isolating the dictionary match. On the other hand, ASCII characters are considered part of the same word, which prevents the policy from triggering.
This behavior is a known limitation in SMG. An enhancement request has been submitted to refine how SMG interprets word boundaries, ensuring consistency in how policies are applied regardless of character encoding.
Once the adjustment is implemented, this article will be updated to reflect the changes and provide guidance for configuration updates.