Vector and Embedding Weaknesses (LLM08)
Risks specific to vector databases, embedding models, and RAG architectures. Includes embedding inversion (recovering source text from embeddings), unauthorized retrieval, and corpus poisoning.
- Rank
- LLM08 of 10
- In the check
- Not cited directly by a question
What it looks like in practice
Three shapes this risk takes in real deployments.
Example 1
An attacker inverts embeddings stored without access controls and recovers training-text.
Example 2
A RAG retrieval returns chunks the requesting user did not have permission to see.
Example 3
A poisoned corpus chunk is retrieved and biases model output.
Controls that close it
The controls the check looks for when this entry applies.
Access control on vector stores aligned to source-document permissions
Embedding-store encryption
Retrieval audit logging
Corpus content review
Where the check cites it
The AI Posture Check cites OWASP LLM Top 10, including this entry, when placing you at Crawl, Walk, Run or Sprint.
The questions cite the control frameworks a regulator would expect you to hold: NIST AI RMF, ISO 42001 and the OWASP LLM Top 10. Your stage on the check is a starting point for an EU AI Act conformity review, not a substitute for one.
Other frameworks the check cites
Score yourself against this framework.
Five questions, each citing its source. You get your stage, your place on the chart and the one move that matters next.
- A few minutes for most people
- Free, from CWS
- Your stage, the chart and the next move, by email or live with an engineer