LLM08: Vector and Embedding Weaknesses
Risk Level: High
Vulnerabilities in vector databases and embedding pipelines used by RAG (Retrieval-Augmented Generation) systems. Attackers can manipulate embeddings to poison knowledge retrieval, bypass access controls in vector stores, or inject malicious content that gets prioritized in similarity searches. Insufficient access controls on vector databases can expose sensitive documents to unauthorized users.
Mitigations
- Implement access controls on vector database collections — enforce user-level permissions
- Validate and sanitize documents before embedding and indexing
- Monitor for adversarial embedding manipulation and anomalous retrieval patterns
- Use metadata filtering to enforce document-level authorization during retrieval
- Regularly audit and re-index vector stores to remove stale or poisoned data