Fully Homomorphic Encryption (FHE) explained

From Caesar’s cypher to Fully Homomorphic Encryption (FHE) – Jeremy Bradley, COO, Zama, explains, in this sponsored article, exactly what FHE is, how it has evolved, what it is now capable of and how far off truly universalised FHE is.

Believe it or not, encryption – a now well-known method of converting information into a code to prevent unauthorised access –  has been around for thousands of years. In fact, one of the earliest examples dates back to 60 BC when Julius Caesar would shift each letter of the alphabet by a fixed number of places to send messages secretly.

Today, of course, Caesar’s cypher – as it was known – would be easily broken with modern techniques. The ancient method did, however, lay the foundation for the significantly evolved encryption techniques of today – with one of the most advanced being Fully Homomorphic Encryption (FHE).

Exactly what is FHE?

The vast majority of those in the tech sphere are fully versed in the complexities of modern encryption methods, but there are still some who aren’t up to speed on exactly what FHE is and how it differs from Homomorphic Encryption (HE).

FHE – which has long been considered a major breakthrough in cryptography – is a specific type of HE that allows complex computations to be performed on encrypted data without needing to decrypt it first. It essentially allows third parties to perform meaningful computations on it without ever accessing the underlying data, meaning data remains secure and private throughout the entire computational process.

While FHE is a HE method, HE is actually a broader term that includes any form of encryption that permits computation on ciphertexts, producing an encrypted result that, when decrypted, matches the result of operations performed on the plaintext. The key differences between FHE and other forms of HE, such as Partially Homomorphic Encryption (PHE) and Somewhat Homomorphic Encryption (SHE), are based on the variety and number of operations that can be performed on encrypted data. FHE supports both addition and multiplication operations on encrypted data, allowing for an unlimited number of computations.

Where might FHE be used?

FHE’s groundbreaking capability has significant implications for data privacy and security both in the public and private sectors, where data is still being stolen because it isn’t encrypted during processing.

Covering fields such as cloud computing, financial services, healthcare, education,  academic and industrial research, and even voting systems, FHE also plays a key role in Secure Multi-Party Computation (SMPC), Intellectual Property Protection, Internet of Things (IoT) security; basically anywhere sensitive data needs to be analysed without exposing the actual information, FHE can step in to secure it. But there are much wider benefits that come with this security.

Take healthcare data analysis, for example. In this scenario, FHE can transform how patient data is utilised while strictly adhering to privacy and regulatory requirements. Medical institutions and researchers can analyse encrypted patient data to identify trends, develop treatment protocols, or conduct large-scale epidemiological studies – all without accessing individual patient records. This not only ensures patient confidentiality and compliance with stringent laws like HIPAA but allows for the secure sharing of medical data across institutions for collaborative research, leading to far better healthcare outcomes.

It’s the same story in the financial sector, where the industry can leverage FHE for a variety of applications. Notably, they can secure data while performing complex analyses like fraud detection, credit scoring, and risk assessment. Here, banks can analyse encrypted financial records, ensuring client confidentiality and data integrity. FHE also facilitates secure data sharing between institutions for anti-fraud purposes, without exposing individual customer data. The ability to compute on encrypted data also opens the door to personalised financial services, all while adhering to privacy regulations.

Where is FHE development at?

For those that have heard of FHE, they may understand its potential and know that it’s a powerful technology, but aren’t too sure what phase of development it’s in.

The concept of FHE actually originated in the late seventies, but the first realisation didn’t come until three decades later. That’s because FHE has faced some challenges in terms of its computational efficiency and performance. Historically, it’s been considered too slow for practical use, but advances in algorithms and computing power have steadily improved its viability for real-world applications. In fact, as of now, not only are developers using FHE to enhance data privacy and security in applications where sensitive data needs to be processed without exposing the underlying information, but FHE is advancing quickly.

The focus is on making it more practical, easier to use and more accessible across various industries. At Zama, for instance, we’ve managed to reduce the bootstrapping time from 20 milliseconds to just 3 milliseconds on CPUs. Additionally, we anticipate a further improvement of 5 to 10 times due to recent breakthroughs in cryptography that we are actively developing. Transitioning to GPUs and FPGAs is expected to bring an additional 5 to 10 times improvement, positioning us to achieve a performance increase of at least 100 times compared to our initial benchmarks at Zama’s inception.

Although these advancements are substantial and sufficient for numerous blockchain and AI applications, we’re still aiming for a 100-fold increase to c not just in blockchain, but across AI and cloud computing applications.

The key to achieving this lies in hardware acceleration. Several leading companies are making strides in this area and are expected to release relevant solutions throughout this year and next. Therefore, we are confident in stating that FHE’s technical challenges are effectively addressed, and by 2026, FHE technology will be widely deployable across various platforms.