iiinova — where your digital self keeps evolving

Niobium @niobiuminc.bsky.social · 02/10/2026

Last week we shared an encrypted genetic risk scoring app that our product manager built from scratch with our FHE skill. His full walkthrough is now up on YouTube. If you’re curious what building with FHE actually looks like, this is a good place to start. #TGIFHE

youtu.be

Building a DNA Analysis App with Fully Homomorphic Encryption (FHE)

YouTube video by Niobium

000

Niobium @niobiuminc.bsky.social · 25/09/2026

Our DNA can tell us a lot about our disease risk, and we can't exactly reset it after a breach. Brad Levergood's latest project, Recessive, calculates those risk scores while the genetic data stays encrypted the entire time. 🧬 Built with our open-source FHE skill for Claude Code. #TGIFHE

github.com

GitHub - blevergood/recessive: Inherited disease risk scored for a whole cohort under fully homomorphic encryption. Every marker is carried into the score. Not one is ever expressed.

Inherited disease risk scored for a whole cohort under fully homomorphic encryption. Every marker is carried into the score. Not one is ever expressed. - blevergood/recessive

020

Niobium @niobiuminc.bsky.social · 18/09/2026

A Rust rewrite of our CLI? We’ll take it. This week’s #TGIFHE comes from Andrew Lewman of Redwoods Research. He built fogrs, an open-source behavioral replacement for the Fog CLI. Same config, same API, and the secret key stays with the data owner while The Fog™ does the compute. Take a look.

code.lewman.com

fogrs: fogrs

031

Niobium @niobiuminc.bsky.social · 17/09/2026

AI agents breached Hugging Face in July. Root on production, 730 MB of private data taken. Breaches like this are why FHE exists: data stays encrypted while it's being computed on. Our CTO Dave Archer explains, by way of a French fleet scuttled in 1942. It’s definitely worth a read.

niobium.co

Scuttled

Toulon, the OpenAI - Hugging Face Fiasco, Torpedo Junction, and FHE’s Convoy Defense

010

Niobium @niobiuminc.bsky.social · 11/09/2026

Andrew Lewman (Redwoods Research) built a language model that answers questions while the question, the model, and the answer stay encrypted. The server never sees any of it in plaintext. Open source: code.lewman.com/llm-fog-fhe/doc/trunk/README.md #TGIFHE

code.lewman.com

llm-fog-fhe: fhe-llm-demo — encrypted tiny QA on the Niobium Fog

022

Niobium @niobiuminc.bsky.social · 10/09/2026

153M driver's license scans are for sale on the dark web, likely leaked by an ID verification vendor. A license check needs to confirm three things: real, valid, and of age. The vendor kept the whole document. On the Fog™, that archive never exists. Steal everything we hold and you get ciphertext.

niobium.co

The 153M License Breach That Was Predictable and Preventable

153 million license scans were offered for sale on a dark-web marketplace. They were collected to answer a question that never needed them. Encrypted compute answers it without building the archive.

011

Niobium @niobiuminc.bsky.social · 04/09/2026

🚨 New #TGIFHE For OpenAI Build Week, our MBA intern Sarah built StealthMatch, an FHE demo, using Codex. Founders can rank investors and grants while sensitive inputs like raise targets and conflict lists stay CKKS-encrypted. Runs on CPU or Niobium FPGAs via The Fog™

devpost.com

StealthMatch

A secure opportunity-matching app for founders, helping to research investors, accelerators, grants, and pilot programs without uploading any confidential facts that determine fit before an NDA.

132

Niobium @niobiuminc.bsky.social · 03/09/2026

Most people think encryption works because nobody knows how it works. On the contrary, all of the algorithms and research are public. The security lives in the math. Our latest FHE 101 post talks through how that math works. niobium.co/press/fhe-10...

niobium.co

Compute on Encrypted Data: The Ideas Behind the Math

Why Modern Cryptography Is Really a Math Story

032

Niobium @niobiuminc.bsky.social · 18/08/2026

A draft last week claimed a quantum algorithm that could break the lattice math behind post-quantum encryption and every practical FHE scheme. Four days later, Gupte, Ragavan, and Zhandry proved it can't recover a single bit of a key. Dave Archer on what happened: niobium.co/press/edward...

niobium.co

Edward Teller, Igniting the Atmosphere, and Why Cryptography Works

A new quantum algorithm briefly raised questions about the security foundations behind lattice cryptography and FHE. Niobium CTO David Archer explains what happened, why the claim deserved scrutiny, a...

020

Niobium @niobiuminc.bsky.social · 14/08/2026

Google published a post on HEIR today, their open-source compiler that makes trained AI models run on encrypted data. Four demos shipped with it. We worked on two: fraud detection with Hardshell, and Kitsune for anomaly detection on encrypted network traffic. Good day for encrypted math.

blog.google

How Google is Making Private AI Practical with Homomorphic Encryption

Today we're excited to showcase HEIR, the latest powerful tool added to our Private Computing Toolkit. HEIR is an open source compiler that unlocks cryptographically-sec…

272

Niobium @niobiuminc.bsky.social · 13/08/2026

Our marketing intern built an FHE health app in three days. VitalVault analyzes sensitive health data without decrypting it, built with Claude Code + Niobium’s open-source tools (and plenty of trial and error). @leiladb.bsky.social wrote up the full build story: niobium.co/press/buildi...

niobium.co

Building VitalVault: Three Days, Zero Coding Experience, One Fully Encrypted Health App

How a marketing intern used Claude Code and Niobium’s open-source tools to build a fully homomorphic encryption (FHE) app that analyzes sensitive health data without decrypting it.

030

Niobium @niobiuminc.bsky.social · 08/08/2026

FHE is easier to understand when you can watch it work. This FHE demo privately determines whether a household qualifies for a peak-hours energy rebate using its 24-hour smart-meter profile. Encrypt the data, compute eligibility on ciphertext, decrypt only on the client. Run it yourself ↓ #TGIFHE

github.com

GitHub - blevergood/offmeter: Private demand-response eligibility scoring under FHE — utilities score households on encrypted smart-meter data without seeing usage. (OpenFHE / Niobium)

Private demand-response eligibility scoring under FHE — utilities score households on encrypted smart-meter data without seeing usage. (OpenFHE / Niobium) - blevergood/offmeter

120

Reposted by Niobium

leiladb.bsky.social @leiladb.bsky.social · 03/08/2026

I have basically no coding experience, but I've been building FHE demo apps with @niobiuminc.bsky.social open source tools and Claude. latest: CipherSpend, a personal finance app that analyzes your spending while every transaction stays encrypted. the server never sees your data.

github.com

GitHub - leila-db/cipherspend-demo

Contribute to leila-db/cipherspend-demo development by creating an account on GitHub.

361

Niobium @niobiuminc.bsky.social · 01/08/2026

Our marketing intern wanted spending insights without letting anyone see her bank transactions. So she built CipherSpend. Full spending analysis on encrypted transactions. Built with our FHE Application Design Agent. Open source: t.co/QYfV4v4yBf #TGIFHE

t.co

http://github.com/leila-db/cipherspend-demo

131

Niobium @niobiuminc.bsky.social · 30/07/2026

Our FHE Design Agent is open source. It takes an ordinary program to a verified encrypted application: privacy model, FHE-friendly redesign, parameter selection, and a test harness that proves outputs match plaintext. Works with any coding agent: github.com/NiobiumInc/n...

github.com

niobium-skills/skills/fhe-application-design at main · NiobiumInc/niobium-skills

A Skill for Anthropic Claude that enables Claude users to more easily create secure computation applications that use fully homomorphic encryption. The initial commit demonstrates an 11% improvemen...

030

Niobium @niobiuminc.bsky.social · 30/07/2026

We gave our FHE Design Agent an XGBoost diabetes model and a privacy requirement. About an hour later ➡️ a working encrypted application. 45,895 records scored under encryption, zero decision flips vs plaintext. The server never saw a single record. Full 5-minute demo:

youtube.com

Niobium FHE Design Agent Demo

YouTube video by Niobium

121

Niobium @niobiuminc.bsky.social · 30/07/2026

Bletchley Park turned codebreaking from an art practiced by a few into a repeatable process, then handed the mechanical parts to machines. We are applying the same lesson to the opposite problem: building encrypted applications. This time the machine works for the encryptor. Read the post:

niobium.co

An AI Agent for Building Encrypted Applications

An FHE AI agent to make encrypted app development fast, rigorous, and open to anyone ready to try.

131

Reposted by Niobium

FHE.org @fheorg.bsky.social · 13/07/2026

Dr. David Archer is Chief Technology Officer of Niobium, maker of cutting-edge hardware acceleration for Fully Homomorphic Encryption. Dr. Archer has over 40 years of experience in developing processor and memory systems, distributed software systems, cryptography, and data-intensive systems.

121