August 15, 2024
5 min read

Bringing Confidentiality to Vector Search with Cyborg and NVIDIA cuVS

Written by
Nicolas Dupont
Related posts
General

Your OpenClaw Agent Remembers Everything. So Would an Attacker.

You don't need to breach the infrastructure anymore. The agent's already inside — with shell access and a plaintext memory of everything it knows about you. OpenClaw changed the threat model. Your vector database hasn't caught up.

General

How NVIDIA's CES 2026 Breakthroughs Accelerate the Encryption-in-Use Mandate

Vector storage and KV caches are the new enterprise breach target. CyborgDB keeps embeddings encrypted during search—built for production AI.

General

Why We Don't Use Homomorphic Encryption (And Why You Shouldn't Require It)

Homomorphic encryption promises secure AI but remains too slow and uncertified for enterprise use. CyborgDB delivers FIPS-compliant encrypted vector search with sub-10ms latency, ready for production today.