A financial institution uses an Amazon Redshift database to store transactional and customer data, including sensitive information such as personally identifiable information (PII). A data analyst needs access to specific columns of the database to perform customer behavior analysis. The institution wants to ensure that sensitive data, such as PII, is not exposed to the analyst while maintaining the integrity of the source data. The solution must avoid duplicating or transforming the data into a separate dataset. What do you recommend to address these requirements with the LEAST implementation effort?