December 17, 2024
We are excited to announce a new capability in BlueXP workload factory to enhance the security and privacy of your generative AI applications. Customers can now filter Personally Identifiable Information (PII) data when building a retrieval augmentation generation (RAG) knowledge base.
The latest update to the Workload Factory GenAI workload introduces a data guardrails capability. This feature enables organizations to create privacy controls for their RAG pipelines. When creating your knowledge base, you now have the option to specify privacy settings that identify and mask PII data discovered in the source documents. With this privacy filter on, the knowledge base, workload factory redacts PII data before the documents are ingested into the knowledge base, preventing PII from being exposed in the embedding and language models.
The PII data guardrail feature is enabled as part of the knowledge base creation workflow in workload factory. To enable the feature, first deploy BlueXP classification service using the BlueXP interface. Once completed, configure data guardrails on the knowledge base during the knowledge base creation process or edit an existing knowledge base to enable the data guardrails.
To learn more, visit the workload factory product page, documentation and announcement blog. To get started now, sign up for workload factory.