The Ministry of Culture, Sports and Tourism and the Korea Copyright Commission held the “2024 AI-Copyright System Improvement Working Group Plenary Meeting” on the 13th at the Seoul office of the Korea Copyright Commission, presided over by Yoon Ho-seong, the first vice minister of the Ministry of Culture, Sports and Tourism. The meeting was organized to consolidate opinions on various issues discussed by the 2024 consultative body (working group) over the past ten months since its inception in February this year.
Over the course of 14 meetings in the past ten months, the working group addressed several issues. Discussions in the learning subcommittee included institutional improvement plans related to the use of copyrighted works included in learning data and the disclosure and methods of AI-learned data. The output subcommittee discussed institutional improvement plans for indicating AI outputs, criteria for determining copyright infringement of AI outputs, and opinions on the protection and registration of AI outputs.
To clear uncertainties related to securing AI data and to obtain quality learning data, the learning subcommittee sought to introduce an exemption regulation for “Text and Data Mining (TDM)” as part of institutional improvement plans for utilizing copyrighted works included in learning data. TDM refers to the act of extracting and analyzing meaningful information on specific topics from various text data for AI learning.
Discussions on the introduction of TDM exemption regulations highlighted the need to clarify the scope of “non-commercial purposes” and noted that if commercial use is not allowed, the regulation may lack effectiveness. However, it was also suggested that minimal TDM exemption regulations are necessary for basic technology development.
There were concerns about whether to introduce TDM exemption regulations including the right holders’ opt-out from AI learning, as refusal may hinder learning high-value data. Nevertheless, right holders, the industry, and technologists were generally positive about adopting the regulations.
The committee members also largely agreed on the need for the disclosure of learning data, suggesting that the methods, subjects, and scope of disclosure be determined through discussion with the industry while considering corporate trade secrets. AI-oriented committee members stated that since learning data includes various information beyond copyrighted works, it should be regulated under the “Basic AI Act,” with supplementary regulation through individual laws like the “Copyright Act” if necessary. They proposed disclosing general information about the data while a third party manages more detailed information, considering issues like trade secrets and personal data leakage, allowing review at the request of the copyright holder.
Regarding the institutional improvement of AI output indication, the output subcommittee suggested drawing from overseas legislative trends like those in the European Union (EU) to define divergent obligations for AI developers, service providers, and users within the AI industry. For instance, imposing the obligation to embed information indicating AI outputs on developers or providers, while preventing users from modifying, altering, distorting, or deleting such information.
Additionally, they remarked that copyright infringement attributable to AI outputs could be judged based on existing elements such as merit and substantial similarity, while asserting a need to categorize and determine responsibility for infringement among AI developers, service providers, and users.
For the protection and registration of AI outputs, it was recommended to guide registrants to specify the portions of AI outputs and human creative contributions during copyright registration. AI-utilized content with human creative input was deemed eligible for copyright registration and protection.
Separately from the consultative body’s activities, the Ministry of Culture, Sports and Tourism has been collecting feedback from key domestic and international AI developers, service providers, right holders, and related parties. A recent “online public opinion survey” conducted from November 4th to December 6th gathered 10,522 responses, reflecting strong public interest in AI-copyright issues.
The ministry plans to announce the “AI-Copyright Policy Direction” based on the findings from the consultative discussions, feedback from AI businesses, right holders, and the general public, and comprehensive research on AI-copyright measures. It also intends to create and distribute guidelines to help the public understand the criteria for determining copyright infringement of AI outputs and the registration of AI-utilized works.
Vice Minister Yoon stated, “With AI technology being widely used in our lives and creative fields, interest in copyright issues has accordingly increased among the public, industry, and right holders. The Ministry will continue to proactively prepare and promote institutional improvement measures to balance the development of the AI industry with the protection of creators’ rights.”