The Culture and Arts Division of the Korea Game Society, the Myongji College Industry-Academic Cooperation Foundation, and OGQ hosted the second 'K-Culture Data Governance Seminar' online on the 20th. Experts from public, private, academic, and research sectors gathered to discuss the 'standards for AI training data sharing and accountability structures'.

Lee Kwon-soo, a team leader at the Korea Culture Information Service, kicked things off by stating that 'AI still struggles to fully grasp the nuances of Korean culture', emphasizing the need for high-quality datasets that capture narrative context. By securing copyright-cleared text, images, and audio, they have been working on 'reference data'—ranging from emotional labeling of traditional patterns to 8K historical costume footage and standardized traditional music. As Lee put it, while the public sector cannot build every single dataset, it must provide what he calls 'the answer key' for AI to reference.

Next, Shin Chul-ho, CEO of OGQ representing the Myongji University Industry-Academic Cooperation Foundation, pointed out the structural tension between AI evolution and copyright protection, proposing an IP Clearing House model as a solution. 'We need a system that manages who owns the rights, who earns what, and how it is all settled,' he argued. He proposed a five-step framework covering rights verification, revenue aggregation, settlement, disclosure, and anti-forgery, insisting that we move away from litigation-heavy reactive measures toward trust-based proactive design.

Professor Lee Eun-jin of Myongji College, who also serves as a board member of the Game Rating and Administration Committee, framed cultural data as a strategic asset that shapes a nation's cognitive framework. 'AI training data isn't just content; it's the bedrock that builds a nation's worldview,' she noted, asserting that cultural sovereignty is essentially a matter of cognitive sovereignty.

She moved beyond the traditional 'Open' policy, suggesting a 'Designed Open' governance model that incorporates purpose-based access control and logged training histories. She advocated for a clear division of labor: the public sector sets the baseline data, while the private sector handles the tech and operations. Crucially, she warned that separating the functions of private aggregators from independent verification infrastructure (the Clearing House) is the only way to prevent the concentration of power.

Gil Se-gi, a center director at the Korea Institute of Sport Science, detailed the roadmap for a sports dataspace, arguing that we need to connect fragmented sports data on a foundation of trust. By leveraging data catalogs, participant authentication, transparent settlement, and a 'compute-to-data' approach, he aims to shift the industry from a production-centric focus to one driven by utility.

The Korea Game Society concluded that this seminar served as a pivotal moment to re-evaluate AI data governance. They have effectively redefined the issue as a national design challenge, built on three pillars: public sector reference data, private sector settlement and rights protection, and a governance structure anchored in cultural sovereignty, all tied together by a robust accountability tracking system.