Baffle, the easiest way to protect sensitive data, is debuting the extension of its data protection platform to pgvector on PostgreSQL. With this release, Baffle expands its data protection and control capabilities to the vector database space, ensuring that enterprises using vector databases for their GenAI apps are secured against potential data leakage.
GenAI adoption has sparked many conversations surrounding effective data security, as vector databases may be responsible for harboring sensitive data within its embeddings. This is supported by research from Gartner, which found that “though data stored in these [vector] DBs is not human readable, it’s still not wholly secure.”
Securing vector database operations—such as similarity search—is a crucial task toward driving GenAI success. With Baffle’s latest innovation, enterprises can utilize the Real Queryable Encryption capability to enforce cryptographic protection that follows regulated data while allowing any, and all, analytical and operational computations on the data—no code changes required.
“For organizations deploying their latest GenAI applications, whether to enhance customer support experience or improve marketing analytics, the sensitive data used needs to be protected in a manner that meets current data security and privacy compliance requirements,” said Ameesh Divatia, co-founder and CEO of Baffle. “Baffle has been able to do this for database and cloud object stores for years, and now we extend this capability to vector databases ensuring that the embeddings of sensitive data are never exposed, while GenAI applications use them.”
Baffle’s enterprise-class data security platform secures data stores for application and GenAI with no-code changes, according to the company. Supporting an array of protection capabilities—such as masking, tokenization, and encryption with role-based access control at the logical database, column-, row-, or field-level—Baffle empowers enterprises to run their GenAI apps with confidence.
To learn more about Baffle’s extension to pgvector on PostgreSQL, please visit https://baffle.io/.