In 2021, the country has successively issued the " Data Security Law of the People's Republic of China " and " Personal Information Protection Law of the People's Republic of China ". The heavy supervision has been implemented, and enterprises will face comprehensive compliance risks. The secret war around data is also heading towards the end. In addition, the Internet dividend has gradually faded, which has caused a number of companies to fall into a growth bottleneck.
According to Gartner's forecast, more than 80% of companies around the world will face at least one privacy-focused data protection regulation by the end of 2023. In this context, as an emerging technology system that can realize "data is not visible", the value of privacy computing is reflected and quickly enters commercial scenarios.
According to iResearch Consulting "China Privacy Computing Industry Research Report", from 2016 to the first quarter of 2022, a total of 55 financing events occurred in China's privacy computing industry, with a cumulative financing amount exceeding RMB 3 billion. In February this year, Lanxiang Zhilian, a privacy computing company from Hangzhou, also announced that it had completed a round A financing of nearly 200 million yuan, which is enough to prove that the privacy computing track still continues the hot market last year.
Although the industry is still in the early stage of large-scale commercialization, and the underlying business logic has not yet been clearly implemented, as policy dividends are gradually released, competition in the privacy computing track may become more intense in 2022. In this regard, Data Yuan interviewed Xu Min, former vice president of Alibaba Cloud , former founder and general manager of Alibaba Financial Cloud, and founder and CEO of Lanxiang Zhilian, and listened to him talk about the value significance of privacy computing and the future trend of the industry.
Blue Elephant Zhilian is a privacy technology and data element ecological service provider, committed to promoting the circulation of data value and releasing the productivity of data elements. Based on rich industry background and data application experience, the company helps customers in industries such as finance, operators, governments, and Internet platforms to better use data elements and enhance business value. It has a top senior team in the industry, with half of the team members coming from Alibaba and Ant Financial.
Privacy calculation - realize data value circulation
In Xu Min's view, privacy calculation is actually an embodiment of the concept of "data value circulation". "There are many ideas for implementing privacy computing, such as multi-party security computing and federal modeling, but the key premise is the same: the original data does not leave the control scope of the data provider. What is really circulated is only the value behind the data." Xu Min explained.
According to the "China Financial Technology Development Report (2021)", the current trend of financial data integration is growing, and the calls for security protection are gradually increasing. As one of the technical means to effectively solve the problem of data privacy protection in the computing process, privacy computing is becoming an important cornerstone of financial technology innovation .
In the past, the rise of big data credit reporting has exposed the weakness of traditional financial institutions in mastering data. Banks' current experience often can only follow the " 28 rule ". Due to the lack of data, it is difficult to verify the risks of most customers, so they can only achieve 80% of their revenue through 20% of their customers. Faced with the massive customers who account for 80% of their "will be willing but not enough", it just reflects a major problem in the financial field, namely the " data island " problem.
"Although financial institutions can build their own multi-dimensional data processing capabilities, there are many optimization points here to obtain and use internal and external data in a safe and compliant manner, and thus promote business development requirements. The original intention of our entry into the privacy computing industry is to improve the data capabilities of traditional financial institutions through technology. If the main task during Alibaba Cloud in the past few years was to enable financial institutions to have the explosive business power of Internet platforms, the main task at Blue Elephant Intelligence in the next few years is to enable more financial institutions to have the ability to apply data like Internet platforms." Xu Min said.
As technology iterative development, traditional financial institutions have gradually realized the gap between themselves and Internet companies. Some financial institutions hired senior operation personnel from the Internet industry to try to replicate the operational capabilities of Internet companies, but in the end the response was mediocre.
In this regard, Xu Min pointed out: "Traditional financial institutions have relatively weak data foundations and cannot effectively support related operational activities. They generally need external data to supplement them, and pay attention to internal and external and cross-institutional data interconnection and mutual collision. This is like a fight between two martial arts masters. No matter who has the best skills, the one who wins in the end is often better internal strength, and data is the 'internal strength' of operation."
But as the saying goes, knowing is easy and hard to do. In commercial practice, privacy calculations often require more than two parties to carry out joint calculations, so the multi-party collaboration characteristics are very obvious. In this process, many privacy computing companies generally adopt multi-directional computing schemes with central nodes, encrypting the data at both ends homomorphically and then placing them in the central node for calculation.
Xu Min told Data Yuan that the above solution has restrictions on the commercial application level, especially among large financial, operators or government and enterprise customers, it is difficult to find a third-party central point that everyone trusts. "Under the traditional model, data must be out of the domain to complete modeling, especially sample data, and the model cannot be protected, which makes it difficult for financial institutions to apply intelligently. Even if the calculation is carried out at the data provider, the data cannot be out of the domain, this will expose the caller's calculation rules and models to a certain extent, which poses great risks to many rule-sensitive enterprises. For example, if the bank's risk control model is exposed, it will mean failure."
In contrast, Lanxiang Zhilian deploys edge computing nodes on both ends of operators and financial institutions based on federated learning modeling, removing untrusted third-party nodes, allowing both ends to complete the encryption calculation and communication process at the same time. Through point-to-point direct connection, while protecting user privacy, the risk of data leakage is fundamentally avoided, and the value of external data can be revealed. At the same time, based on fully understanding the actual scenario needs of the financial industry, Lanxiang Zhilian has continuously innovated at the product level and developed many practical functions, such as supporting the deployment of the model on the business side, protecting the data of the data and the sensitive models of the business side in model prediction.
GAIA——Financial-level privacy computing product
iResearch Consulting research shows that the number of financial institutions invested in privacy computing in 2022 is about twice or more than twice that in 2021. The demand side of privacy computing has exploded, which has obviously attracted many players to flock to the track, making the industry increasingly crowded.
Facing the continuous emergence of various products on the market, Xu Min roughly divided them into three categories: the first category is laboratory-level products, that is, prototype products with preliminary functions, but cannot adapt to the diversity, complex environment and performance requirements of the enterprise business; the second category is enterprise-level products, which are widely used and can cover multiple industries; and the third category is financial-level products.
The overall IT level in the financial industry is relatively high, and its degree of digitalization is also ahead of many industries. The requirements for privacy computing products are naturally more stringent. "Financial institutions pursue high availability and high performance, but also have high requirements for product security and manageability. For example, in the current business scenarios involving Blue Elephant Intelligent Technology, financial institutions often need to obtain results within one or two hundred milliseconds. Even under the conditions of a multi-institutional network with complex diversity, or in the case of a complex data network composed of multiple institutions, they must maintain such performance requirements, which is undoubtedly a great test for privacy computing service providers." Xu Min introduced.
, which aims to empower the financial industry, has adapted to the strict application requirements of financial-grade products and launched GAIA, which focuses on the application of privacy computing technology in the financial field. It is reported that the GAIA product series of Lanxiang Zhilian mainly includes four major sectors.
one-stop federal learning modeling platform GAIA·Cube supports the full process of visual interactive multi-party secure computing AI R&D capabilities from asset discovery, privacy interception, feature engineering, model training to model security deployment.
multi-party security computing platform GAIA·Edge is a new generation of decentralized basic security data sharing platform based on multi-party security computing technology, which is upgraded by integrating technologies such as big data and blockchain, and provides trusted data asset exchange, industry model prediction, and distributed multi-party security privacy data intelligent application products and solutions.
Privacy Computing Alliance Sharing Platform GAIA·Edge-X is a data security sharing platform that integrates data without leaving the library through multi-party security computing technology, providing a data security sharing platform that integrates data asset full life cycle management , multi-mode networking, data asset transactions and measurement billing, real-time privacy computing and query services.
GAIA data element circulation platform, and uses three major systems such as the data element market platform, data element support platform, and privacy computing capability platform to build a data element circulation platform. In the underlying technical architecture, by integrating cryptography and AI big data technology, it realizes "outdoor circulation can be controlled, ciphertext can be calculated, data can be confirmed, privacy can be audited, and calculation can be measured."
According to Xu Min, unlike some privacy computing products developed based on open source products, GAIA completely adopts the overall technology independently developed, including a self-developed underlying algorithm library. While achieving high performance, high security and other enterprise application requirements, it can also targeted improvement and optimization for specific scenarios.
It is reported that the Blue Elephant Intelligent Privacy Computing Platform GAIA has been widely used in the special attention list sharing scenarios and the anti-fraud alliance platform scenarios respectively. Taking the anti-fraud alliance platform scenario as an example, Blue Elephant Intelligence cooperated with regulatory agencies to introduce the capabilities and data of China Mobile, China Telecom , China Unicom and relevant public security departments. Through two major technical support, multi-party security computing and federal learning, and efficiently coordinated with the "card break" (bank card and mobile phone card) action, effectively building a local anti-fraud barrier.
In addition, as a financial-grade privacy computing product, GAIA has a complete security system, and its security computing engine combines cryptography, key management, MPC protocol, and security operator architecture layered; and it also comes with a key management system (KMS), which is used for the full life cycle management of managed keys and certificates.
mentioned earlier that data operations need to be closely integrated with customer business, and for financial products, marketing is often a big problem. In the process of finding new traffic and acquiring customers in public domain traffic, the approval risk control process usually intercepts 50-90% of the incoming traffic, resulting in a surge in customer acquisition costs of financial product .
In this regard, Xu Min pointed out that through the GAIA platform connecting the three major operators and Toutiao and other delivery channels, financial institutions can achieve high-responsiveness and accurate identification of low-risk customers across the domain, and complete customer screening on the traffic end, thereby improving the customer acquisition efficiency of the entire process.
Based on this privacy computing power, at the end of last year, Lanxiangzhilian cooperated with Industrial and Commercial Bank of China and China UnionPay to promote the circulation and utilization of the data value of ICBC and UnionPay through multi-party data security sharing and joint modeling, breaking the barriers to information exchange and effectively balancing data privacy protection and value mining. Thanks to this, the project was also successfully selected as the benchmark case of "Xinghe" in 2021 big data, further highlighting the technical capabilities of Blue Elephant Intelligent.
"Privacy Computing +" - an ideal form of industry evolution
The development of emerging technologies is always full of opportunities and challenges, and so is privacy computing. According to iResearch Consulting's "Privacy Computing Development Cycle Insight Matrix", the current privacy computing industry is still in the infancy of technology. With the continuous increase in technological expectations, the market is increasingly affirming the technical value of privacy computing, and capital popularity is also continuing to heat up.
. According to statistics from China Institute of Information and Communications Technology , in 2018 and 2019, almost no domestic privacy computing applications entered the deployment stage, but the proportion of privacy computing products entering the deployment stage in 2020 rose to 38%, and continued to rise to 48% in 2021. It can be seen that all kinds of players are entering the market one after another, and competition within the track is also intensifying.
It is worth noting that iRe predicts that the active development of platform construction by leading companies in the industry will become the next trend in the privacy computing track, and closely followed by the tail companies in the industry will also invest in platform construction. In words, the "platformization" wave of the track seems to be revealed.
For this, Xu Min told Data Ape that platformization may be a result, not a target. All of Lanxiang Zhilian's current work is focusing on customer value and business goals, constantly adding bricks and tiles to the underlying technology or data operation layer, accumulating capabilities, thereby enhancing the overall value of the privacy computing industry. As for whether the final business form is a platform or what positioning of each party in the platform, it has gradually evolved.
In his opinion, in the future, privacy computing will become the underlying infrastructure of digital economy , building a solid data application foundation for all industries. Based on this, how to create new industrial changes through the data connectivity capabilities granted by privacy computing is the opportunity in the privacy computing track.
"Privacy computing cannot be called an industry in essence, but should be regarded as a basic technology. This is like the current Internet is no longer the name of a single industry. Only Internet finance , Internet e-commerce, etc., the ' Internet + track' has become an industry. Therefore, if one day the privacy computing track disappears and the 'privacy computing + track' emerges and prospers, the privacy computing industry can truly enter the ideal form." Xu Min explained.
. Against this background, the privacy computing industry has begun to differentiate. At present, there are many types of market participants in privacy computing, and the capabilities, technical paths, and development strategies of different manufacturers are also different. Tong Ling, founder and chairman of Lanxiang Zhilian, said that Lanxiang itself is positioned as a promoter of digital transformation of and digital economy in various industries, and takes pragmatism as its own development path.
Take open source as an example, Xu Min showed a cautious attitude towards this: "If the current size and ability still do not reach the level of operating the open source community , it will be difficult to achieve the goal expected by open source." It can be seen that compared to "open source for open source", Lanxiang Zhilian is more willing to focus on polishing the product itself and regard open source as a "natural".
adheres to this pragmatic business logic. Up to now, Lanxiang Zhilian has reached cooperation with three major operators, China Mobile, China Telecom, and China Unicom, providing all-round cooperation and support in terms of technology, data operations and customer resources; and cooperated with Industrial and Commercial Bank of China and China UnionPay to provide more effective data foundation for the expansion of inclusive financial services, effectively balanced data privacy protection and value mining, and assisted ICBC to bring small and micro financial services to thousands of households. In addition, it has also formed cooperation with dozens of financial or other industry customers, including Bank of Communications, Industrial Bank , Shanghai Bank , Nanjing Bank , Hangzhou Bank , New Network Bank , Guotai Junan , etc., and based on privacy computing technology, improve data capabilities and improve institutions' marketing and risk control capabilities. Looking to the future, Xu Min believes that with the development of privacy computing in the next two years, with the development of privacy computing, there will be four changes in the industry: first, data circulation based on privacy computing will become the mainstream; second, commercial applications based on privacy computing and data circulation will be more prosperous; third, data infrastructure and privacy computing technology will be deeply integrated; fourth, enterprise data compliance requirements are higher.Blue Elephant Intelligent is also actively "evolution" based on the upcoming changes in the industry - Blue Elephant Intelligent's self-positioning "privacy technology and data element ecological service provider" seems to be demonstrating Blue Elephant Intelligent's path to self-evolution.
French thinker Michel de Montaigne once said: "If the traveler does not determine the destination port of his navigation, then any wind direction will not be a good wind." Blue Elephant Zhilian, whose mission is to "promote the circulation of data value and release the productivity of data elements", seems to have also carried this concept, and through linking product technology, data operations and business operations, it promotes the healthy, stable and rapid development of the data element market, allowing data elements to play a greater value in industries such as finance, and bring more beautiful changes to society.
article:Waichi / Data ape
