During the document tokenization process, utilize a language model to detect potential PII and notify users if any is found, so it can be handled appropriately. Consider using a similar approach for Financial Sensitive Information (FSI) and other Controlled Unclassified Information (CUI).
Just to inform, so they know its there but does not stop the usage