Fully Homomorphic Encryption (FHE) is transforming the future of secure data processing. In a recent expert interview, Erin Hales — a final-year PhD researcher at Royal Holloway and an FHE specialist — shared her insights on the evolving landscape of homomorphic encryption. This conversation offered valuable perspectives on the technical, practical, and research-driven dimensions of FHE, particularly around challenges like noise management, real-world applications, and future directions.
Erin Hales: Journey Into FHE
With a background in mathematics, Erin transitioned into cryptography during her PhD, eventually focusing on FHE after attending a Microsoft bootcamp. There, she was introduced to the SEAL homomorphic encryption library, which sparked her interest in the field. Her research has since delved into how noise — an intrinsic element of FHE — affects the correctness and security of encrypted computations.
The Noise Challenge in FHE
One of the most critical technical aspects of FHE is noise management. Noise is intentionally added during encryption to ensure security, but it grows with each operation performed on the ciphertext. If it grows too much, it leads to decryption failures, where the decrypted result becomes meaningless.
Erin’s research focuses on improving methods to estimate and manage this noise, a key challenge because exact values of noise cannot be known without compromising security. Her work contributes to various FHE libraries and helps inform community standards for practical FHE implementation.
Contributing to FHE Standardization
In the absence of ISO/IEC standards for FHE, the cryptographic community has developed informal standardization practices. Erin is involved in these collaborative efforts, helping shape best practices across academia and industry. This kind of agile, community-driven standardization is helping FHE mature more rapidly and adapt to practical use cases.
Is FHE Really Too Slow?
FHE has long had a reputation for being impractically slow. Early implementations took 30 minutes just to perform one bootstrapping operation. However, Erin emphasizes that things have drastically improved: modern schemes can now complete bootstrapping in fractions of a second.
Ongoing advancements in:
- Algorithm optimization
- Efficient software libraries
- Specialized hardware for FHE
…are making real-time FHE more feasible. Erin stresses that while performance still lags compared to unencrypted computation, the trade-off for privacy is often worth it.
Real-World Applications of FHE
Erin highlighted several compelling use cases where FHE is already being deployed or is under development:
- Microsoft Edge Password Monitor: Uses FHE to compare user passwords against breach databases without exposing plaintext data.
- Fitness Tracking Apps (like Strava): FHE enables location data processing without revealing sensitive geographic trails. This is especially critical in cases like military base privacy, where data misuse could pose security risks.
These examples showcase how FHE enables meaningful computation while keeping user data private — a crucial distinction in today’s surveillance-prone digital ecosystem.
The Future of FHE: Integration and Specialization
Looking ahead, Erin envisions the future of FHE in two main directions:
- Application-Specific Optimization: Startups are tailoring FHE schemes for machine learning as a service (MLaaS), private search, and financial analytics.
- Hybrid Privacy Architectures: Combining FHE with technologies like MPC (Multi-Party Computation), Differential Privacy, and Zero-Knowledge Proofs to achieve stronger guarantees or better performance.
This collaborative mindset can help overcome the limitations of standalone FHE and unlock new possibilities in privacy engineering.
Security and Ongoing Research
Standard FHE schemes currently satisfy IND-CPA (Indistinguishability under Chosen Plaintext Attack). However, there’s growing interest in IND-CPAD — a slightly stronger model — as researchers explore how to bolster FHE’s cryptographic robustness. Achieving IND-CCA (Chosen Ciphertext Attack resistance) remains difficult due to FHE’s malleability, a feature that allows computations on ciphertexts but also prevents straightforward security enhancements.
Erin believes the solution may lie in cross-domain collaboration, using tools from other areas of cryptography to strike the right balance between functionality and security.
Final Thoughts
Erin Hales brings clarity and enthusiasm to a field often seen as overly technical or inaccessible. Her work on noise estimation, practical implementations, and community standards is helping push FHE closer to widespread adoption. As hardware, software, and theoretical tools evolve, FHE is poised to become a foundational technology in secure computing.
We love to share our knowledge on current technologies. Our motto is ‘Do our best so that we can’t blame ourselves for anything“.