Podcast: Fully Homomorphic Encryption All Around w/ Kurt Ruhloff, Duality Technologies

This week on the FHE Onchain Podcast, I had Kurt Ruhloff from Duality Technologies joining me — someone who’s been around FHE long before most of us even heard of it. We talked about the early days of homomorphic encryption, how it evolved from pure research to real-world use cases, and the role of open source in making that happen.

Kurt’s Background

Kurt comes from an engineering and defense background. Before co-founding Duality Technologies, he spent years working on secure computing systems, mainly for DARPA, the US government’s advanced R&D agency. That’s also how he got involved with FHE — back in 2010 — at a time when it was still fresh out of Craig Gentry’s first paper.

He was part of DARPA’s early programs trying to make FHE real — not just on paper, but in code.

Why DARPA Cared About FHE So Early

DARPA’s motivation was pretty straightforward: secure cloud computing before the cloud became a mainstream thing. The idea was: what if sensitive data could be stored and computed on remotely without ever decrypting it? Not just privacy for privacy’s sake — but protecting data in use, in untrusted environments.

DARPA also had a rough thesis: cryptography usually takes 20 years to go from academic papers to real-world adoption (see RSA or ECC). They wanted to accelerate that for FHE.

Duality’s Role

Most people in the space know Duality for their contributions to OpenFHE — one of the main open-source FHE libraries today. But beyond the technical work, a lot of Kurt’s focus has been on standardization efforts, collaborating with NIST, ISO, and other bodies to push for interoperability and transparency.

That open-source-first culture in FHE isn’t accidental. It’s deeply tied to the nature of the problem — trust, collaboration, and security can’t exist in closed systems, especially when the goal is secure multi-party computation.

Where FHE Adoption Is Happening Today

Kurt’s view on FHE adoption is pretty grounded. It’s slow, but steady — and most of the early traction is in highly regulated industries:
→ healthcare,
→ finance,
→ government.

That’s where sensitive data already exists, where privacy rules are strict, and where data sharing is broken today because of legal friction.

In these environments, FHE doesn’t compete with plaintext compute — it competes with legal contracts, NDAs, and expensive data-sharing agreements. If encrypting data lets you skip months of legal work, slower compute becomes less of a problem.

Confidential AI

We also talked about confidential machine learning — one of the big topics in the FHE world lately. Kurt sees fine-tuning as a very promising direction: encrypting sensitive data to fine-tune existing ML models without leaking information. Not running massive models fully encrypted — but targeted, specific use cases where privacy unlocks new kinds of collaboration.

Confidential Healthcare

Healthcare is another area Kurt is bullish on — especially for rare disease research, where data fragmentation is a huge blocker. Patients are spread across geographies, and privacy regulations prevent pooling their data easily. FHE (combined with other PETs) could enable collaboration across countries while respecting local privacy laws.

FHE & Web3

Kurt’s view on FHE in Web3 is cautiously optimistic. He sees clear potential around things like private transactions, confidential trading strategies, and risk analysis — especially in cross-border, decentralized environments.

But he’s also realistic that FHE adoption in Web3 will follow the same pattern as elsewhere: slow integration into existing workflows, rather than massive new paradigms overnight.

What’s Next for FHE?

Kurt’s take is simple: for FHE to scale, it has to feel boring. Meaning: users and developers shouldn’t have to care about it. It should integrate into existing frameworks (Python, NumPy, machine learning libraries) without changing how people work.

There’s still technical work to do — better programmability, compiler support, and optimizing for specific hardware. But the trajectory is clear: FHE is moving from an academic curiosity to an actual building block for privacy-preserving applications.

Book Recommendation from Kurt

Arrowsmith by Sinclair Lewis — a novel about science, research, and what it really means to innovate in society. Written almost a century ago, but still hits home for anyone working on hard, long-term problems.