AI RESEARCH

VectorSmuggle: Steganographic Exfiltration in Embedding Stores and a Cryptographic Provenance Defense

arXiv CS.LG

ArXi:2605.13764v1 Announce Type: cross Modern retrieval-augmented generation (RAG) systems convert sensitive content into high-dimensional embeddings and them in vector databases that treat the resulting numerical artifacts as opaque. Major vector- products do not provide native controls for embedding integrity, ingestion-time distributional anomaly detection, or cryptographic provenance attestation.