Vector DB Component
The Vector Database stores embeddings that represent text, images, and other data in a high-dimensional space, enabling semantic search and retrieval augmentation for the LLM.
Component Overview
The Vector Database is a critical component for Retrieval Augmented Generation (RAG) applications, storing vector representations of data that can be efficiently queried to provide relevant context to the LLM.
Security issues in the Vector DB component can lead to data poisoning attacks and exploitation of weaknesses in how embeddings are generated, stored, or retrieved. These vulnerabilities can result in manipulated model outputs or unauthorized access to sensitive information.
Related Vulnerabilities
Data and Model Poisoning
Data poisoning occurs when pre-training, fine-tuning, or embedding data is manipulated to introduce vulnerabilities, backdoors, or biases.
Vector and Embedding Weaknesses
Weaknesses in how vectors and embeddings are generated, stored, or retrieved can be exploited by malicious actions (intentional or unintentional) to inject harmful content, manipulate model outputs, or access sensitive information.