Internet

“We Don’t Send Data to Personal Information Servers”… AI Developed by Former Samsung Security Expert Selected for the Ministry of SMEs and Startups’ Deep Tech TIPS Program

Personal AI to Receive 1.5 Billion Won in R&D Funding for Up to Three Years CEO Lee Jae-young, Former Member of SamsungElectronics’ Security Team, to Lead Galaxy Biometric Authentication AI Training on Smartphones… Reduces the Risk of Leaking Raw Personal Data Efforts Underway to Develop Customized AI for Insurance Claims, Credit Scoring, and More

Kim Hyun-ah
2026-11-22 09:56:52
[Edaily Reporter Kim Hyun-ah ] Technology is being developed that trains artificial intelligence (AI) to tailor services to individuals without sending personal information to external servers. This approach analyzes users’ financial transactions and health records directly on their smartphones to provide AI services. It is drawing attention because it can reduce the risk of large-scale data breaches compared to existing methods that centralize personal information in a single location.

The company developing this technology is Personal AI, Inc., led by CEO Lee Jae-young, a former member of the security team at SamsungElectronics(005930) ’s Mobile Research Lab. The company is applying his experience in developing security technologies—such as fingerprint and iris recognition for Galaxy smartphones—to AI personalization technology.

Personal AI announced that it has been selected for “Deep Tech TIPS,” a technology startup support program run by the Ministry of SMEs and Startups, and will receive 1.5 billion won in R&D funding for up to three years. This project will be led by S&P Lab, the company’s in-house research institute.

Source: Personal AI
, Inc.

Personal Information Stored on Smartphones… AI Training Also Performed On-Device
The company’s competitive edge lies in its “on-device AI” technology, which trains AI on users’
devices
without collecting personal information on corporate servers.

Generally, for AI to provide personalized services, it must analyze users’ consumption history, health information, and other data. If personal information is transmitted to external servers during this process, there is a risk that data from multiple Saramins could be leaked all at once due to hacking or other security breaches.

Personal AI is developing a method that trains small language models (sLLMs) on the device itself while keeping the original personal data on the smartphone.

For example, even if the AI analyzes credit card transaction history to manage spending or provides personalized services based on health records, the detailed transaction records and original health information are not sent outside the device.

The company also applies “federated learning” technology, in which multiple smartphones train the AI using their own data and share only the training results rather than the original personal information. It combines this with “local differential privacy” technology, which protects information by making it difficult to identify individuals based on the training results.

In addition, the plan is to develop technology that uses zero-knowledge proofs and blockchain to verify that each smartphone has trained according to predetermined rules. The goal is to ensure the integrity of the AI training process without disclosing personal information.

Of course, simply not sending the original data to a server does not eliminate all security risks. Since there is a possibility that the smartphone itself could be hacked or that some information could be inferred from the shared training results, additional protective technologies are necessary.

Lee Jae-young, CEO of Personal AI. Photo: Newsis

CEO Lee Jae-young, Former Head of Security for Samsung Galaxy CEO
Lee Jae-young is a mobile security and privacy expert who previously served as a senior researcher in the Security Team at SamsungElectronics’ Mobile Research Institute. He was responsible for technologies and policies related to fingerprint and iris recognition, as well as lock screen security, for Galaxy smartphones.

He also holds certifications as a Certified Information Systems Security Professional (CISSP), Certified Information Privacy Professional/Europe (CIPP/E), and Certified Ethical Hacker (CEH). He currently serves as an adjunct professor in the Department of AI Technology Management at Kyung Hee University, where he teaches courses on AI, blockchain, and MyData.

CEO Lee co-founded S&P Lab in 2019 with developers from SamsungElectronics and Samsung SDS. The company was later renamed Personal AI, and S&P Lab is now used as the name of the company’s in-house research institute.

The company has been developing “my:D,” a MyData platform that manages personal information on smartphones. It also applied zero-knowledge proof technology to the Ministry of National Defense’s “Jangbyeong e-Eum” service, which verifies eligibility for welfare benefits without disclosing personal information.

Expanding AI Business to Include Insurance Claims
and Credit
Scoring…
Through this Deep Tech TIPS project, Personal AI plans to extend its personal information protection technology to AI agents.

AI agents are services that go beyond simply answering users’ questions to actually handling tasks on their behalf. The company plans to develop AI that utilizes individuals’ health, financial, and lifestyle data to support out-of-pocket insurance claims, expense management, and the trading of personal data usage rights.

The development goals include improving personalization performance by at least 10 percentage points compared to existing systems, maintaining at least 90% of the performance achieved through central server-based learning, and achieving an Agent AI task success rate of at least 80%. These are not yet achieved results but rather goals that must be validated through future research and development.

The company is also expanding its business areas to include healthcare, alternative credit scoring, and personal data trading. This involves analyzing non-financial data from smartphones—such as that of Saramins with limited financial transaction histories—to create supplementary credit scoring indicators, or analyzing patients’ health information within medical devices and providing only consented results to medical professionals.

Personal AI announced that, following its return to profitability in 2024, it recorded revenue of 4.32 billion won in 2025. By being selected for the Deep Tech TIPS program—in addition to the existing TIPS and Post-TIPS programs—the company has also laid the groundwork for expanding its technology development.

CEO Lee Jae-young stated, “Personal data breaches stem not only from individual companies’ lax security but also from the structural issue of centralizing personal data on a single server,” adding, “We will complete an AI engine capable of personalization and verification without sending raw data externally, and expand into healthcare, finance, and data trading businesses.”

In an era where AI utilizes even individuals’ financial and health information, ensuring both personal data protection and service performance is a critical challenge. Attention is focused on whether Personal AI can protect personal data within smartphones while delivering performance comparable to server-based AI.

Economy

Corporation

IT·Science

Economy

Restaurants and IT Services Up for Sale… Demand for Acquisitions of Mobility and Food & Beverage Companies [M&A Deal Board]

Every week, in collaboration with LISTING, an M&A matching platform for small and medium-sized enterprises, we report on newly listed assets and acquisition demand in the domestic M&A market. By exami…
2026-11-22 10:00:07

Corporation

“What Makes It Different?”… CU Sells 1.3 Million “Suhyangmi” Ready-to-Eat Meals in Just Two Months

The convenience store chain CU is targeting consumers’ palates with ready-to-eat meals that emphasize authentic rice flavor. Ready-to-eat meals made with “Suhyangmi,” a premium rice variety known for …
2026-11-22 09:42:31

IT·Science

“We Don’t Send Data to Personal Information Servers”… AI Developed by Former Samsung Security Expert Selected for the Ministry of SMEs and Startups’ Deep Tech TIPS Program

Technology is being developed that trains artificial intelligence (AI) to tailor services to individuals without sending personal information to external servers. This approach analyzes users’ financi…
2026-11-22 09:56:52