Skip to content
The Cyborg Log

Notes on Secure AI.

Releases, interviews, and deep dives on building AI that respects the data it touches.

Pinned
CYBORG
Partnerships·Nov 5, 2025

Blueprint: Secure RAG Workflows Without Sacrificing AI Performance

The Cyborg Enterprise RAG Blueprint, now on build.nvidia.com, pairs CyborgDB encryption-in-use with NVIDIA Nemotron and NeMo Retriever for secure, high-performance enterprise RAG.

Nicolas Dupont
CYBORG
Videos·Oct 8, 2025

OpenSSL Conference 2025: End-to-End Encrypted Vector Databases

Nicolas Dupont's OpenSSL Conference 2025 talk on building end-to-end encrypted vector databases — keeping embeddings, queries, and retrieval encrypted at every step.

Nicolas Dupont
CYBORG
Interviews · Videos·Oct 2, 2025

NYSE Wired Interview with Cyborg CEO

NYSE Wired sits down with Cyborg CEO Nicolas Dupont to talk confidential AI, encrypted vector search, and how regulated industries can finally adopt AI safely.

Nicolas Dupont
CYBORG
Videos·Jul 7, 2025

Confidential Computing Summit 2025: The Missing Piece for Confidential AI

Nicolas Dupont's talk from the 2025 Confidential Computing Summit on why confidential vector databases are the missing layer in any serious confidential AI stack.

Nicolas Dupont
CYBORG
Technical Explainer·Jun 13, 2026

Tenant Isolation Keeps Breaking in Vector Databases

On June 12, ChromaDB disclosed four tenant isolation CVEs in a single coordinated advisory — the clearest signal yet that application-layer isolation is not an architectural solution.

Charlcye Mitchell
CYBORG
Videos·Jun 1, 2026

CyborgDB Teaser

A short look at what we're building at Cyborg this quarter — confidential AI, encrypted vector search, and what's coming next.

Nicolas Dupont
CYBORG
Partnerships·May 5, 2026

Cyborg Partners with Austin AI to Deliver End-to-End Secure AI in Production

Cyborg and Austin Artificial Intelligence announce a partnership to deliver end-to-end secure AI in production, combining Cyborg's encrypted vector search with Austin AI's secure inference platform.

Cyborg Team
CYBORG
Interviews · Videos·Apr 1, 2026

RSA 2026: Why Vector DBs Are an AI Security Blind Spot

Cyborg CEO Nicolas Dupont explains why vector databases have quietly become one of the most exploitable surfaces in enterprise AI — and what it takes to actually secure them.

Nicolas Dupont
CYBORG
Interviews · Videos·Mar 31, 2026

RSA 2026: Your Vector DB Might Be a Weak Link in Your Security Strategy

Information Security Media Group catches up with Cyborg CEO Nicolas Dupont at RSAC 2026 to talk about the state of confidential AI and why data privacy in AI pipelines is now a boardroom conversation.

Nicolas Dupont
CYBORG
Interviews · Videos·Mar 25, 2026

GTC 2026: Nicolas Dupont on Securing the Vector Database

Nicolas Dupont breaks down the unique security challenges of vector databases and how Cyborg is solving them with encryption that works at query time — without sacrificing performance.

Nicolas Dupont
CYBORG
Articles·Mar 4, 2026

The Right Vector Database Is a Threat Model Question, Not a Scale Question

Choosing between in-process vector DBs and CyborgDB isn't about scale or ops overhead — it's about your threat model, and most comparisons get this wrong.

Charlcye Mitchell
CYBORG
Articles·Mar 2, 2026

AI Has a Memory Problem. We're Fixing It.

Every enterprise AI runs on a vector memory layer that everyone is racing to make faster — and almost nobody is asking what happens when that memory gets compromised.

Charlcye Mitchell
CYBORG
Articles·Feb 9, 2026

Your OpenClaw Agent Remembers Everything. So Would an Attacker.

OpenClaw stores conversations and personal data as plaintext embeddings — a semantic index of your life that any attacker who breaches the host can reconstruct.

Charlcye Mitchell
CYBORG
Articles·Jan 5, 2026

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

NVIDIA's CES 2026 keynote made encryption-in-use feasible at AI scale — and made it mandatory for any regulated enterprise centralizing data into vector memory.

Charlcye Mitchell
CYBORG
Technical Explainer·Dec 17, 2025

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

Homomorphic encryption is elegant in theory but 1,000–10,000x slower in practice and lacks FIPS certification — making it unsuitable for production AI workloads today.

Charlcye Mitchell
CYBORG
Articles·Dec 15, 2025

How Developers Are Rethinking AI Security: Lessons from the CyborgDB Hackathon

Phase 1 hackathon winners across fintech, healthcare, and enterprise show what AI architecture looks like when encryption is treated as a foundation, not an afterthought.

Charlcye Mitchell
CYBORG
Articles·Nov 24, 2025

When Authentication Isn't Enough: Lessons from Milvus CVE-2025-64513

A critical Milvus auth bypass (CVSS 9.3) exposes a deeper truth: authentication controls who gets in, but only encryption controls what they can do once inside.

Charlcye Mitchell
CYBORG
Articles·Nov 18, 2025

AI Espionage: Why Vector Database Security Just Became Mission-Critical

Anthropic disrupted the first AI-orchestrated cyber espionage campaign — and centralized vector databases are now the highest-value target in the new threat landscape.

Charlcye Mitchell
CYBORG
Partnerships·Nov 5, 2025

Blueprint: Secure RAG Workflows Without Sacrificing AI Performance

The Cyborg Enterprise RAG Blueprint, now on build.nvidia.com, pairs CyborgDB encryption-in-use with NVIDIA Nemotron and NeMo Retriever for secure, high-performance enterprise RAG.

Nicolas Dupont
CYBORG
Articles·Nov 3, 2025

The Cyborg Hackathon: Build Real-Time Encrypted AI

$10,000 in prizes across two phases — we're challenging developers to build production AI systems on fully encrypted vector search. Registration closes November 23.

Charlcye Mitchell
CYBORG
Partnerships·Oct 15, 2025

Cyborg & Redpanda: Secure Streaming Pipelines for Enterprise AI

Cyborg joins Redpanda's 'Hello, Agent' podcast to talk through how to build secure, real-time streaming pipelines that keep enterprise AI data encrypted end-to-end.

Cyborg Team
CYBORG
Releases·Oct 14, 2025

Starting Our Open Source Journey

We're open-sourcing CyborgDB, starting with our Python, TypeScript, JavaScript, and Go client SDKs today — with the service layer to follow later this year.

Charlcye Mitchell
CYBORG
Videos·Oct 8, 2025

OpenSSL Conference 2025: End-to-End Encrypted Vector Databases

Nicolas Dupont's OpenSSL Conference 2025 talk on building end-to-end encrypted vector databases — keeping embeddings, queries, and retrieval encrypted at every step.

Nicolas Dupont
CYBORG
Interviews · Videos·Oct 2, 2025

NYSE Wired Interview with Cyborg CEO

NYSE Wired sits down with Cyborg CEO Nicolas Dupont to talk confidential AI, encrypted vector search, and how regulated industries can finally adopt AI safely.

Nicolas Dupont
CYBORG
Technical Explainer·Sep 4, 2025

Vector Embeddings Are Not One-Way Hashes

Embeddings aren't anonymized — inversion attacks can reconstruct original text and images from vectors, meaning your embedding store deserves the same rigor as raw data.

Nicolas Dupont
CYBORG
Partnerships·Aug 20, 2025

Secure AI Inference Pipelines with Cyborg & Protopia AI

Cyborg and Protopia AI partner to lock down end-to-end AI inference pipelines — protecting prompts, embeddings, and retrieval from exposure at every step.

Cyborg Team
CYBORG
Technical Explainer·Aug 14, 2025

Your Enterprise AI Application Is a Security Time Bomb

Centralizing enterprise data into a vector database makes AI smarter — and creates a single high-value breach target whose fallout dwarfs any traditional database leak.

Nicolas Dupont
CYBORG
Videos·Jul 7, 2025

Confidential Computing Summit 2025: The Missing Piece for Confidential AI

Nicolas Dupont's talk from the 2025 Confidential Computing Summit on why confidential vector databases are the missing layer in any serious confidential AI stack.

Nicolas Dupont
CYBORG
Interviews · Videos·Jun 20, 2025

Ciphering Intelligence: The Future of AI Encryption

Nicolas Dupont joins Robotics & AI Infrastructure Leaders to discuss how encryption is evolving to meet the demands of modern AI systems — and why confidentiality must be built in from the ground up.

Nicolas Dupont
CYBORG
Technical Explainer·May 8, 2025

What is Confidential AI? (And Why It Matters)

Confidential AI keeps data encrypted through every stage of inference — combining trusted execution environments and confidential vector databases to close the plaintext gap.

Jessica Wanczyk
CYBORG
Releases·Apr 15, 2025

Introducing CyborgDB

Meet CyborgDB — the first Confidential Vector Database that keeps embeddings end-to-end encrypted at rest, in transit, and even during inference.

Nicolas Dupont
CYBORG
Partnerships·Aug 15, 2024

Bringing Confidentiality to Vector Search with Cyborg and NVIDIA cuVS

Cyborg teams up with NVIDIA cuVS to bring GPU-accelerated vector search to confidential workloads — encrypted embeddings, compliant inference, no performance compromise.

Nicolas Dupont
CYBORG
Releases·Nov 2, 2023

Hello Stealth, Goodbye Beta

After 12 months and 40,000 beta testers, Stealth is generally available — bringing zero-knowledge end-to-end encrypted cloud storage, team files, and link sharing.

Nicolas Dupont
CYBORG
Technical Explainer·Jan 6, 2023

How Stealth's Encrypted Search Works

A look under the hood at the cryptographic indexing, hashing, and trapdoor functions that let Stealth search end-to-end encrypted data without ever decrypting it.

Alexandre Helle
CYBORG
Technical Explainer·Dec 5, 2022

What is End-to-End Encryption?

A plain-English breakdown of end-to-end encryption — what makes it different from encryption-in-transit, and why it's the gold standard for protecting your data.

Alexandre Helle
CYBORG
Technical Explainer·Nov 17, 2022

What is Encrypted Search?

Encrypted search has long been the holy grail of cryptography — keep data unreadable to the server while still letting users find what they're looking for.

Alexandre Helle
CYBORG
Articles·Oct 31, 2022

Why We Care About Data Privacy

When you waive digital privacy you're not just giving up data — you're handing tech companies the leverage to shape what you buy, watch, and ultimately think.

Nicolas Dupont