Cloaked AI

Overview

Cloaked AI is a security solution from IronCore Labs that enables application-layer encryption for vector embeddings, allowing secure storage and querying of AI data in vector databases like Pinecone, Qdrant, and Weaviate.

Security Threats Addressed

AI applications rely on vector embeddings to power search and recommendations, but these data-rich vectors introduce new security and privacy risks:

• Data reconstruction attacks from embeddings
• Unauthorized access to vectorized data
• Manipulation of AI applications
• Insertion of bias and inappropriate content

Encryption Solutions

Application-Layer Encryption (ALE)

The best way to secure embeddings is using application-layer encryption (ALE), which means you encrypt the data before sending it to a vector database or to an index file using something like FAISS.

Searchable/Queryable Encryption

Searchable encryption (SE), also called queryable encryption, allows users to search encrypted data without decrypting it. This capability is especially useful when:

• Data privacy is a concern
• Using cloud storage
• Operating in shared environments

Property-Preserving Encryption

Property-preserving vector encryption lets developers query a vector database if and only if they have the right key. Without the key, they can't meaningfully query it.

Integration

Works with popular vector databases:

• Pinecone
• Qdrant
• Weaviate
• FAISS

Best Practices

• Encrypt data both in transit and at rest
• Implement strict access control measures
• Monitor who can view, modify, or manage the database
• Use TLS encryption for data in transit

Pricing

Enterprise pricing through IronCore Labs.