Introducing Fully Homomorphic Encryption (FHE): Exploring its exciting applications, limitations, and recent developments that have fueled its popularity.
When I first heard about “Fully Homomorphic Encryption” (FHE), I was intrigued by the blockchain space’s tendency to give buzzy concepts long names. Over the years, we’ve come across many buzzwords sweeping the industry, the most recent of which is “Zero Knowledge Proofs” (ZKPs).
After some investigation and exploring new companies that are building products using FHE, I noticed a landscape filled with a whole new set of tools. In the coming months and years, FHE could become the next big technology sweeping the industry, just like ZKPs did. Companies are leveraging recent advances in various areas of cryptography and cloud computing to pave the way to a robust, data privacy-preserving future. The question is not if we can achieve this, but when, and I believe FHE could be a key enabler for progress in data privacy and ownership.
Over the next few weeks, I will be diving deeper into learning more about FHE and examining its limitations, potential, and applications. I will be sharing my findings in a series of posts exploring different aspects of the conversation around FHE. This week, I will introduce the technology and discuss why it has been getting a lot of attention lately. Many people in the industry are talking about it, including Kyle Samani[4] from Multicoin Capital, who said:
“FHE is the holy grail of cryptography. Over time, FHE will reshape the structure of all computation, both in web2 and web3.”
The key to solving the problem is to understand what we mean by “homomorphism.” Tracing its roots, homomorphism has its roots in mathematics, where it is defined as a mapping between two algebraic structures of the same type that preserves core components.
If you prefer a more practical definition like me, a fundamental principle behind the mathematics is that two groups do not need to be exactly the same to have the same core properties. For example, imagine two boxes of fruit:
Box A contains small fruits.
Box B contains large fruits.
Despite the different sizes of the individual fruits, squeezing a small apple and a small orange in box A will produce a mixed juice that tastes the same as squeezing a large apple and a large orange in box B. Squeezing juice to produce the same taste is akin to preserving core components between the two boxes. Assuming our main concern is the same taste, then it doesn't matter which box we squeeze the juice from, because the amount of juice is not what we care about. The two boxes are equivalent in the area of interest (taste), so the differences between them (size and amount) have no effect on their main function (producing a specific juice taste).
By analogy with homomorphism, we capture its two main features:
Mapping: We establish a connection between the two boxes, where each small fruit in box A corresponds to a large version in box B. So, a small apple in box A corresponds to a large apple in box B, and so on.
Preservation of operations: If squeezing two small fruits in box A produces a juice of a specific taste, then squeezing their corresponding large versions in box B should produce the same taste. Despite the difference in size and amount of the juice, the "taste characteristic" is preserved.
To tie all this back to the central theme of this article, fully homomorphic encryption[6] is a specific method of data encryption that enables one to perform computations on encrypted data without revealing the original data. In theory, analysis and computations performed on encrypted data should produce the same results as if performed on the original data. With FHE, we establish a 1:1 relationship between the data in the encrypted dataset and the data in the original dataset. In this case, the core component that remains is the ability to perform any computation on either dataset and produce the same results.
Against this backdrop, many companies have taken precautions to protect user data and maintain differential privacy. Companies rarely store data in the cloud or in their databases in raw, unencrypted form. Therefore, even if an attacker gains control of a company’s servers, they still have to bypass the encryption to read and access the data. However, data is no longer interesting when it is merely encrypted and sitting idle. When companies want to perform analysis on data to derive valuable insights, they have no choice but to decrypt the data. Once decrypted, the data becomes vulnerable. However, with end-to-end encryption, FHE becomes extremely useful because we no longer need to decrypt the data to perform analysis; this is just the tip of the iceberg of possibilities.
A key consideration is whether companies should be allowed to read and store our personal information. The standard response from many people to this is that companies need to see our data in order to provide us with a better service.
If YouTube doesn't store data like my viewing and search history, the algorithms can't do their best to show me the videos I'm interested in. Therefore, many people believe that the trade-off between data privacy and getting a better service is worth it. However, with FHE, we no longer have to make this trade-off. Companies like YouTube can train their algorithms on encrypted data and produce the same results for the end user without violating data privacy. Specifically, they can homomorphically encrypt information like my viewing and search history, analyze it without ever seeing it, and then show me the videos I'm interested in based on the analysis.
FHE is an important step towards a future in which our data is no longer a valuable commodity that we voluntarily give away for free to organizations.
Fully homomorphic encryption (FHE) when applied correctly is a breakthrough for all sectors that store user data. We are looking at a technology that could change our entire attitude towards data privacy, and the extent to which companies are willing to infringe.
Let’s start by examining how FHE is reshaping data practices in the healthcare industry[7]. Many hospitals keep private records of patients stored in their databases, which they must keep confidential for ethical and legal reasons. However, this information is valuable to outside medical researchers, who can analyze the data to infer important insights into diseases and potential treatments. A major obstacle that has slowed research progress is maintaining absolute confidentiality of patient data when outsourcing data to researchers. There are many ways to anonymize or pseudo-anonymize patient records. But they are not perfect and may reveal too much information about someone, making them identifiable, or not reveal enough information about their case, making it difficult to gain accurate insights about a disease. Using fully homomorphic encryption (FHE), hospitals can encrypt patient data, making it easier to protect patient privacy in the cloud. Medical researchers can perform computations and run analytical functions on encrypted data without compromising patient privacy. Since there is a one-to-one mapping between encrypted and original data, the results obtained from encrypted datasets provide real insights that can be applied to real cases. FHE can quickly advance the healthcare industry.
Another exciting application of FHE is artificial intelligence (AI) training. Currently, the field of AI faces privacy issues that prevent companies from accessing many extensive datasets used to perfect AI algorithms. Companies training AI must choose between using limited public datasets, paying large amounts of money to purchase private datasets, or creating datasets, which is challenging for smaller companies with fewer users. FHE should address the privacy issues that prevent many dataset providers from entering this market. As a result, improvements in FHE could lead to an increase in the number of datasets available for training AI. This will make AI training more financially accessible and more sophisticated as the diversity of available datasets increases.
If fully homomorphic encryption (FHE) is indeed a game-changer for modern big data, why haven’t we seen more real-world applications yet?
Despite years of discussion and research into FHE, in reality, it’s very difficult to implement in practice. The core challenge lies in the computational power required to perform FHE. Fully homomorphically secure datasets can produce the same analytical results as their original data form. This is a challenging feat that requires a lot of computational speed and power, much of which is impractical to implement on existing computers. Operations that typically take seconds on the original data can take hours or even days on a homomorphically encrypted dataset. This computational challenge creates a self-perpetuating cycle where many engineers put off pursuing FHE projects, slowing their development and limiting the realization of their full benefits.
One specific example of a computational problem engineers face in FHE is how to address “noise errors[8]”. When performing computations on homomorphically encrypted datasets, many engineers generate excess noise or errors each time they perform a computation. This is tolerable when only a few computations need to be performed, but after multiple analyses, the noise can become so noticeable that the original data becomes unintelligible. The data is all but lost.
Just as generative AI[9] was once considered limited and primitive before becoming mainstream, fully homomorphic encryption (FHE) is heading in the direction of similar progress. Many industry leaders, even beyond the blockchain space, have come together to organize a large amount of FHE research and development. This has led to several recent industry developments that are driving a compelling narrative of progress for this technology.
In March 2021, Microsoft, Intel, and the Defense Advanced Research Projects Agency (DARPA) agreed to launch a multi-year program[10] to accelerate the development of fully homomorphic encryption (FHE). The program, called Data Protection in Virtual Environments (DPRIVE), marks a major step forward for FHE. It shows two industry giants specializing in cloud computing and computer hardware joining forces to address data privacy issues. They launched the program to build computers and software capable of managing the speed of FHE calculations and to develop guidelines for accurately implementing FHE and guarding against data leaks that could result from incorrect use.
As part of the DPRIVE program, engineers have begun addressing the “noise error” mentioned earlier, exploring ways to reduce the noise level to preserve the original data. One promising solution is to design large arithmetic word length[11] (LAWS) data representations. While traditional computer processors (CPUs) typically use 64-bit word lengths, engineers are developing new hardware capable of processing word lengths of 1024 bits or more, employing LAWS. This approach is effective because research has shown that longer word lengths directly affect the signal-to-noise ratio. Simply put, longer word lengths produce less noise with each additional computation in FHE, allowing more computations to be performed before reaching a data loss threshold. By building new hardware to address these challenges, engineers involved in the DPRIVE program have significantly reduced the computational load required to perform FHE.
To speed up computations, approaching the goal of making FHE 100,000 times faster, the DPRIVE team embarked on an ongoing journey to design new data processing systems that go beyond the capabilities of traditional processing and graphics units. They developed a new multiple-instruction, multiple-data[12] (MIMD) system capable of managing multiple instructions and data sets simultaneously. MIMD is akin to building a new highway instead of using existing roads that are underequipped to accommodate the volume of traffic required for the fast, real-time computations of FHE.
What’s interesting about the DPRIVE program is the extensive use of “parallelism[13]” in computer mathematical computations. This enables developers to perform multiple large-number calculations at the same time. You can think of parallelism as deploying a group of mathematicians to work on different parts of a huge mathematical problem at the same time, rather than having them do their work one by one. While performing multiple calculations at the same time helps solve problems quickly, the computers must be air-cooled to prevent overheating.
In September 2022, a year and a half after launching the program, Microsoft, Intel, and DARPA announced[14] that they had successfully completed the first phase of the DPRIVE program. They are currently working on the second phase of DPRIVE. Intel also launched[15] its own fully homomorphic encryption toolkit to provide developers with tools to facilitate faster fully homomorphic encryption in the cloud. Intel designed this toolkit to ensure compatibility with the latest advances in data processing and computing. It includes special features tailored specifically for lattice cryptography, integration that works seamlessly with Microsoft Seal, samples of fully homomorphic encryption schemes, and technical documentation to guide users.
Google’s Private Join and Compute[16] open source library provides developers with tools for multi-party computation (MPC). This computational approach allows parties to gain shared insights by merging their different data sets without exposing the original data to each other. Private Join and Compute combines cryptographic techniques from FHE with private set intersection (PSI) to optimize data confidentiality practices. PSI is another cryptographic approach that allows parties with different data sets to identify common elements or data points without revealing their data. Google’s approach to advancing data privacy is not just focused on FHE; it prioritizes broader MPC concepts by integrating FHE with other influential data practices.
The increasing availability of well-respected open source libraries for fully homomorphic encryption is noteworthy. However, the picture becomes even more compelling when we observe well-respected companies experimenting with these libraries in their operations. In April 2021, Nasdaq, a prominent stock exchange and global capital markets technology entity, incorporated FHE into its operations. Leveraging Intel’s FHE tools and high-speed processors, Nasdaq is tackling financial crime through anti-money laundering efforts and fraud detection by using fully homomorphic encryption to identify valuable insights and potential illegal activity in data sets containing sensitive information.
In addition to the research and development conducted by the companies mentioned above, several other companies have recently received significant funding for initiatives focused on fully homomorphic encryption (FHE).
Cornami[18], a large technology company, is widely credited for pioneering the development of scalable cloud computing technology designed for fully homomorphic encryption. They are engaged in numerous efforts to create computing systems that support FHE more efficiently than traditional CPUs. They also direct initiatives aimed at protecting encrypted data from the threat of quantum computing. In May 2022, Cornami announced a successful Series C funding round led by Softbank, raising $68 million, bringing its total funding to $150 million. Zama is another company in the blockchain industry that is building open-source fully homomorphic encryption tools that developers can use to build exciting applications using FHE, blockchain, and AI. Zama has developed the Fully Homomorphic Ethereum Virtual Machine (fhEVM) as one of its products. This smart contract protocol enables transaction data on-chain to remain encrypted while being processed. Developers exploring the use of the Zama library for various applications have been impressed by its performance, even in complex use cases. Zama successfully closed a $42 million Series A round in February 2022, led by Protocol Labs, bringing its total funding to $50 million.
Fhenix[22] is also an emerging project that is bringing FHE to the blockchain. Their goal is to expand FHE applications beyond confidential payments, opening the door to exciting use cases for FHE in areas such as decentralized finance (DeFi), bridging, governance voting, and Web3 gaming[23]. In September 2023, Fhenix announced that it had successfully closed a $7 million seed round, led by Multicoin Capital and Collider Ventures.
For years, fully homomorphic encryption (FHE) has been an idea that promises strong end-to-end encryption, heralding a future of strong data privacy. Recent developments are beginning to transform FHE from a theoretical dream into a practical reality. While companies are racing to be the first to pioneer a strong, fully functional version of FHE, many are collaborating to collectively tackle the complexities of this powerful technology. This spirit of collaboration is evident through the implementation of various cross-team initiatives and the development of open source libraries that integrate with other libraries.
Based on my research, the discussion around FHE appears to be far-reaching. In the coming weeks, I’m excited to dive in and share more insights from my research on FHE. Specifically, I can’t wait to explore the following topics:
Emerging applications of FHE.
The interplay between zero-knowledge proofs (ZKPs) and FHE.
The interplay between zero-knowledge proofs (ZKPs) and FHE.
Integrating FHE with Private Set Intersection (PSI) to advance secure multi-party computation (MPC).
New companies like Zama and Fhenix are pioneering developments in the field of FHE.
Arampatzis, Anastasios. “What are the latest developments in fully homomorphic encryption?” Venafi, February 1, 2022, venafi.com/blog/what-are-latest-developments-homomorphic-encryption-ask-experts/.
Arampatzis, Anastasios. “What is Fully Homomorphic Encryption and How to Use It.” Venafi, April 28, 2023, venafi.com/blog/homomorphic-encryption-what-it-and-how-it-used/.
“Building Hardware for Continuous Data Protection.” DARPA, March 2, 2020, www.darpa.mil/news-events/2020-03-02.[25]
Cristobal, Samuel. “Fully Homomorphic Encryption: The Holy Grail of Cryptography.” Datascience.Aero, January 7, 2021, datascience.aero/fully-homomorphic-encryption-the-holy-grail-of-cryptography/.
“Fully Homomorphic Encryption: What It Is and Why It Matters?” Internet Society, March 9, 2023 Hunt, James. “FHENIX Raises $7M in Seed Round Led by Multicoin Capital.” The Block, The Block, September 26, 2023, www.theblock.co/post/252931/fhenix-seed-multicoin-capital. “Intel® Fully Homomorphic Encryption Toolkit. ” Intel, www.intel.com/content/www/us/en/developer/tools/homomorphic-encryption/overview.html#gs.fu55im.[28] Accessed October 8, 2023.
“Intel and Microsoft Collaborate on DARPA Program.” Intel, March 8, 2021, www.intel.com/content/www/us/en/newsroom/news/intel-collaborate-microsoft-darpa-program.html#gs.ftusxq.[29]“Intel to NASDAQ Advances in Homomorphic Encryption Research.” ” Intel, April 6, 2021, www.intel.com/content/www/us/en/newsroom/news/xeon-advances-nasdaqs-homomorphic-encryption-rd.html#gs.6mpgme.[30]
Johnson, Rick. “Intel Completes DARPA DPRIVE Phase One Milestone for a Fully Homomorphic Encryption Platform.” Intel, September 14, 2022, community.intel.com/t5/Blogs/Products-and-Solutions/HPC/Intel-Completes-DARPA-DPRIVE-Phase-One-Milestone-for-a-Fully/post/1411021.
“Microsoft Seal: Fast and Easy-to-Use Homomorphic Encryption Library.” ” Microsoft Research, January 4, 2023, www.microsoft.com/en-us/research/project/microsoft-seal/.[31]
Paillier, Pascal, PhD. “Fully Homomorphic Encryption: The Holy Grail of Cryptography.” Business Age, March 9, 2023, www.businessage.com/post/fully-homomorphic-encryption-the-holy-grail-of-cryptography[32]
Samani, Kyle. “The Dawn of On-Chain FHE. ” Multicoin Capital, September 26, 2023, https://multicoin.capital/2023/09/26/the-dawn-of-on-chain-fhe/
“What is Fully Homomorphic Encryption?” Inpher, April 11, 2021, https://inpher.io/technology/what-is-fully-homomorphic-encryption/
Fully homomorphic encryption is often hailed as the holy grail of cryptography. This article explores the application prospects of FHE in the field of artificial intelligence and points out the current limitations. It also lists some projects dedicated to using fully homomorphic encryption for AI applications in the field of encryption.
JinseFinanceAlthough the crypto market in 2024 is not as turbulent as in the past, there are still some new technologies trying to mature.
JinseFinanceAlthough the crypto market in 2024 is not as turbulent as in the past, there are still some new technologies trying to mature, such as the topic we are going to talk about today: "FHE/Fully Homomorphic Encryption".
JinseFinanceFHE fully homomorphic encryption is the next generation of privacy protection technology that is about to emerge and is worth our planning.
JinseFinanceIntroducing Fully Homomorphic Encryption (FHE): Explore its exciting applications, limitations, and recent developments that have fueled its popularity.
JinseFinanceWith Zama completing $73 million in financing, the fully homomorphic encryption ecosystem has emerged.
JinseFinanceIsomorphic binding is one of the core technologies used by the Bitcoin layer asset issuance protocol RGB++. Through this technology, RGB++ can solve various problems encountered by the RGB protocol and give RGB more possibilities.
JinseFinanceZero-knowledge proofs (ZKP) are clearly useful for improving scalability and privacy in web3, but are hampered by reliance on third parties to handle unencrypted data.
JinseFinanceBarry Silbert was an early investor in the crypto industry, getting his start all the way back in 2012.
ZoeyUser experience is a determining factor in DeFi adoption; fortunately, wallets may be the bridge needed to overcome this hurdle.
Cointelegraph