Reflecting Field Challenges:
Streamlined Risk Assessment and Processing System

The Personal Information Protection Commission announced on March 30 that it will completely revise the "Pseudonymous Information Processing Guidelines" to reflect the rapidly changing data utilization environment, including the transition to artificial intelligence (AI).


The pseudonymous information system is designed to enable the safe use of data—which is essential for AI development—without the risk of personal information breaches. By pseudonymizing data so that individuals cannot be identified, it allows data to be used for research purposes without the consent of the data subjects.


Personal Information Protection Commission logo.

Personal Information Protection Commission logo.

View original image

According to the Commission, this revision comprehensively incorporates the challenges encountered in the field since the introduction of the pseudonymous information system.


First, a standardized risk assessment system has been newly established. Until now, the pseudonymous information system had different judgment criteria depending on the organization or individual in charge, causing confusion in the field. In particular, the lack of standardized criteria often led to different judgments on the same issue, depending on the subjective interpretation of each reviewer.


As a result, the new guidelines move away from a method that examines every complex risk factor. Instead, risk levels are now determined based on "who is utilizing the data" and "under what environment the data is processed." Specifically, internal use is considered "low risk," while provision to third parties is classified as "medium risk" or "high risk" depending on whether the processing environment can be controlled. Furthermore, the guidelines allow flexibility to adjust risk levels upward or downward, taking into account the specific circumstances of each case and internal policies of institutions.


Risk assessment flowchart of the revised pseudonymous information processing guidelines. Personal Information Protection Commission

Risk assessment flowchart of the revised pseudonymous information processing guidelines. Personal Information Protection Commission

View original image

The pseudonymous information processing system will also be streamlined. The review methods and required documentation will be differentiated according to risk level, so that lower-risk cases can be handled more quickly and easily. The number of required document formats will also be reduced from 24 to 10.


This addresses concerns that the complex review procedures and extensive documentation requirements during the pseudonymization process have been burdensome for field practitioners. There have been many cases where excessive review procedures and documentation were uniformly required for all matters, regardless of the actual risk involved.


For example, if pseudonymous information is used within the same organization to compile service usage statistics, there is no new risk arising from external provision, so it is classified as "low risk." In this case, the process can be handled by the person in charge without a separate review committee, and only minimal documentation is required.


Revised Contents of the Pseudonymous Information Processing Guidelines for AI Development. Personal Information Protection Commission

Revised Contents of the Pseudonymous Information Processing Guidelines for AI Development. Personal Information Protection Commission

View original image

The operational standards for the pseudonymous information system are also being adapted to the AI era. Under the revision, it will be permitted to set and review "expandable purposes" within a similar scope in advance, allowing the same pseudonymous information to be used repeatedly for similar purposes. As AI continues to learn from data and improves its performance, the criteria for setting processing periods have also been made more flexible, so that pseudonymous information can be used continuously for the period necessary for AI service development and advancement.


The revision also takes into account the growing prevalence of multimodal AI models that handle not only text but also videos and images. For large-scale unstructured data such as videos and images, it is practically difficult to conduct a full inspection to verify proper pseudonymization. Therefore, the guidelines now allow for sample inspections of selected data, which enhances the efficiency of data processing.


Additionally, the guidelines have been divided into a main section (system overview) and a supplementary section (practical processing guide) to make them easier to understand and use. The main section is structured to make concepts and procedures easily accessible to everyone, while the supplementary section provides practical content such as pseudonymization techniques, safety measures, and instructions on filling out forms. Usage scenarios and frequently asked questions have also been greatly expanded to address situations commonly encountered in the field.



Song Kyunghee, Chairperson of the Personal Information Protection Commission, stated, "Up until now, the pseudonymous information system has had a high entry barrier in the field due to complex procedures and conservative operation. By listening to the difficulties and opinions from the field from the ground up and completely overhauling the guidelines based on actual risk levels, we expect this to serve as a turning point where the safe and effective use of pseudonymous information will increase dramatically in the accelerating AI transformation environment."


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing