On May 20, the personal information including the names, schools, and mobile phone numbers of more than 10,000 teachers selected for the training course for the application of AI digital textbooks was leaked by the Ministry of Education. This incident proves that the Ministry of Education is unqualified to be in charge of the development and implementation of AI digital textbooks, which require the collection of a vast amount of personal information. Open Net Korea urges the Ministry of Education to postpone the development of AI digital textbooks until it has completed its privacy measures.

AI digital textbooks are not digitized versions of hardcover textbooks, but rather an “educational platform” used by students, teachers, and parents. Defined as “learning courseware (learning course + software) designed from the student’s perspective” that provides “learning diagnosis and analysis by AI” and “customized learning that reflects individual learning levels and speeds,” AI digital textbooks consist of an AI digital textbook portal provided by the public (Ministry of Education and Korea Education and Research Information Service (KERIS)), subject-specific AI digital textbooks provided by private (AI digital textbook developers), and a learning data hub. The portal functions as a conduit for students and teachers to access digital textbooks, a bookshelf where students can check their own subjects and timetables, and an integrated dashboard where students can receive personalized learning support, parents can check their children’s learning status, and teachers can design lessons and guide students. The information provided in the unified dashboard is the result of analyzing learning data as students interact with textbooks and learn. The learning data is sent to the learning data hub, where it is analyzed by a private developer and sent to the portal’s integrated dashboard. That means that the digital textbook is a software platform accessible to students, parents, and teachers, as well as private developers. In the “Guidelines for the Development of AI Digital Textbooks” (hereinafter the guidelines) published in 2023, the KERIS stated that the digital textbook is to “contribute to the improvement of the overall quality of education by fulfilling the “platform’s” function of enabling data-driven education decision-making at the national level.

<AI Digital Textbook Service Configuration Diagram>
From the Guidelines for the Development of AI Digital Textbooks

As a platform, the key enabler for AI digital textbooks is the learning data, or personal information, that students produce while they study or learn. Learning time, content performance, content metadata, achievement, access time, formative assessment achievement, question and answer and so on will consist of the collected learning data. According to the guidelines, the collected personal information of students will be used for the following analytical purposes:

1. To assess students’ current understanding by analyzing their learning characteristics, and to recommend suitable learning content and paths that align with their individual abilities and goals.
2. To deliver appropriate learning materials by comprehensively analyzing students’ learning patterns, including their interests, preferences, and proficiency levels.
3. To enhance students’ conceptual understanding through AI tutor functionalities.
4. To offer customized services, such as additional learning elements, by analyzing students’ learning patterns and activities.

The guidelines state that the data collected by the AI digital textbooks should be kept to a minimum. However, given the recent incident of teacher personal information leakage by the Ministry of Education, we are concerned about the possibility of leakage and misuse of children’s personal information when AI digital textbooks are commercialized. Children’s personal information collected in an educational environment is directly related to digital rights and requires more careful attention than adults’ personal information. In a session titled “Privacy and Surveillance Issues in AI Digital Textbook Policy,” hosted by Open Net at the 3rd Summit for Democracy on March 20, Prof. Jeong Hyun-sun highlighted that “the information included in the AI digital textbook service, particularly the analysis of ‘learning attitude,’ ‘interests,’ ‘preferences,’ ‘learning activity status,’ and ‘academic emotion,’ is sensitive information. This analysis, which involves an AI system inferring human emotions in an educational environment, is classified as high-risk under the EU AI Act and should be handled with special care.” Son Ju Eun, the discussant of the session, added that “data collected on children’s identity, learning activities, location, communication, emotions, health, and social relationships can be combined to accurately identify a specific child, even when pseudonymized or anonymized.” According to the guidelines, digital textbooks are also expected to collect fingerprints and facial images to facilitate login for learners, teachers, and parents to the digital textbook portal.

The distribution and usage of AI digital textbooks inevitably mean that government departments will collect the personal information of all children subject to compulsory education through private companies. Therefore, experts have advised that a system for monitoring and supervising personal information protection should be established in parallel with the development of the textbooks, and the process should be transparent and open to the public. In the above presentation of the session, Prof. Jeong suggested that “concrete and effective enforcement ordinances and regulations should be established for the protection of children’s personal information, along with supervision and regulatory mechanisms for developers.” She also emphasized that “research studies, public hearings involving students, parents, teachers, and the general public, as well as securing budget and personnel, are urgently required.” Additionally, she recommended that “research, legislation, and supervision should be carried out by an independent institution with secured personnel and budget.” The guidelines identify the KERIS as the main body for technology inspection and management but do not mention the specific system (i.e. the department in charge, the content of the work, the number of required personnel, etc.). To address public concerns about the potential leak of teachers’ personal information, the Ministry of Education and the KERIS should proactively release a comprehensive plan for the management, monitoring, and supervision of personal information collected by digital textbooks.

The AI digital textbooks will transform the role of the Ministry of Education. As a personal information processor, MOE is responsible for monitoring and overseeing personal information and information security systems. The amount of personal information collected by AI digital textbooks will be incomparable to the amount of information that the Ministry of Education and frontline schools have been collecting until now. As the amount of personal information grows, the MOE’s responsibilities will need to be strengthened. Nevertheless, the ministry’s sloppy handling led to the leak of sensitive information of more than 10,000 private tutors. This incident raises fundamental doubts about the ministry’s ability to monitor and supervise the privacy of AI digital textbooks. OpenNet Korea calls on the Ministry of Education to put a moratorium on the development of AI textbooks until it has established solid privacy measures and transparently disclosed them to reassure school sites.

Korean version text