Exploring Future Tasks for Copyright Protection in the Era of AI Generation

Korea Copyright Protection Agency Holds the fourth “2025 Copyright Protection

Future Forum”

The Results of Systemic Improvement

“Copyright Protection at a New Turning Point”

In his opening remarks, President Park highlighted the recent revision to the Copyright Act, passed by the National Assembly’s Culture, Sports and Tourism Committee by saying, “The copyright protection system is at a new turning point. With this revision, the authority to order the blocking of illegal websites, previously held by the Korea Media and Communications Commission, shifts to the Minister of Culture, Sports and Tourism. Such authority will be subject to review by the KCOPA Copyright Protection Deliberation Committee.”

In addition, he noted that the introduction of an “emergency blocking” system is also being pursued, allowing the Minister of Culture, Sports and Tourism to immediately block access to pirated content whenever the rapid proliferation of pirated content poses a risk of irreparable damage if left unchecked.

President Park said, “The Copyright Protection Future Forum, which has been held for the past two years, serves as an important platform for designing copyright protection governance.” In commemoration of KCOPA’s 10th anniversary next year, he expressed his commitment to further strengthening the protection system in response to technological and industrial structural changes.

In his welcoming speech, Attorney Yim Sanghyuk (Shin & Kim LLC.) praised the forum by saying, “This forum stands as a key platform for exploring legal and policy solutions surrounding copyright, moral rights, and fair use in the era of AI.” He also stated that the Ministry of Culture, Sports and Tourism (MCST) has been developing the fair use guidelines for AI learning through its “copyright-AI working group,” with a public briefing slated soon before announcing the final draft.

First Presentation

The Emergence of Moral Rights Issue

in the Era of Generative AI

During the first presentation, Attorney Hwang Jiwon of Shin & Kim LLC. addressed the recent increase in international AI-related litigation under the theme of “Generative AI and the Protection of Moral Rights.” She emphasized, “As AI learns from human creations and generates similar results, the potential for infringement of moral rights looms as a key issue beyond the existing copyright infringement.”

Attorney Hwang pointed out that the discussion of moral rights related to AI should be considered from two perspectives: ▲ whether moral rights can be recognized for the output of generative AI; and ▲ whether AI outputs can infringe on the moral rights of existing authors. Through examples, she then explained how moral rights, such as the right to publication, the right to attribution, and the right to maintain integrity, can be affected by the AI learning and generation process.

Examples that she presented included cases where TDM immunity was not recognized in the German case of GEMA v. Open AI, as well as cases where AI cover songs and voice cloning can infringe on the rights of publicity and the personal interests of performers. Moreover, she predicted that Korea’s lack of a discovery system poses extreme difficulties for copyright holders in detecting unauthorized use of copyrighted works by AI operators. Lastly, she emphasized, “In an AI-driven content ecosystem, institutional improvements that provide incentives for creators remain essential for sustainable creation.”

Second Presentation

The Direction of Next-Generation AI Technology

and the Changing Copyright Environment

During the second presentation, Professor Bae Juho of Hankuk University of Foreign Studies examined key trends in AI technology between 2024 and 2025. He noted that AI performance has recently improved significantly, now evident even to the public, and outlined emerging experimental evidence for a recursive self-improvement (RSI) method, in which generative AI relearns its own output to improve performance. He explained how conventional wisdom once held that retraining self-output would result in quality degradation. However, recent “Aha moment” observations, where significant performance spikes occur even in small models, now serve as the foundation for the evolution of major models, such as GPT and Claude, into thinking models.

He then proposed test-time computing, where multiple models perform inference in parallel and only the optimal results are reflected in learning, as another key technology trend. This technology’s nonlinear performance improvements, fueled by massive computational infrastructure, prompted him to argue that U.S. and China’s AI investments now carry the character of an “unstoppable arms race.”

Professor Bae expanded the discussion to real-world applications, emphasizing the rapid advancement of physical AI operating in real-world environments, including autonomous driving, service robots, and domestic robots. He also predicted that this could lead to changes in labor and industrial structures. He said, “The key competency in creative and education fields lies in how AI is utilized, not whether it is used,” while proposing a restructuring of the education system centered on guiding individual students’ curiosity and exploration.

He pointed out that AI operates in a “continuous learning structure,” which combines pre-training, post-training, and inference processes, arguing that the existing method of dividing it into “learning stage vs. use stage” and regulating it will soon reach its limits. He emphasized that future legal, policy, and ethical discussions will hinge on how far unauthorized use will be considered and what technological and institutional boundaries can be drawn.

Comprehensive Proposal

The Need for Administrative Blocking

and Data System Reform

Professor Lee Kyuho of Chung-Ang University summarized the key points of this year’s discussions and presented future policy directions through the comprehensive proposal “2025 Copyright Protection Future Forum Achievements and Future Tasks.” Professor Lee said, “The greatest achievement of this year’s forum is the fact that experts regularly gathered to exchange opinions,” emphasizing the need for a sustained discussion structure based on collective intelligence.

Professor Lee explained that various blocking methods for illegal and harmful websites, including websites, IP addresses, and DNS, already exist, and if the amendment is passed, it is highly likely that the system will transition to an administrative blocking system, with the Minister of Culture, Sports, and Tourism at the helm of blocking access. Administrative blocking proves a highly effective means of copyright protection within Asia, he emphasized, having introduced Thailand’s first case of blocking illegal websites and Malaysia’s administrative blocking system.

Among future tasks, he proposed ▲ shifting from static blocking to dynamic injunction, ▲ introducing live blocking capable of real-time response, and ▲ establishing a rapid response system that matches the speed of AI proliferation. He also proposed expanding the scope of fair use to include moral rights and considering exemption from the obligation to cite sources in fair use and parody areas. He further emphasized the importance of exempting attribution violations from criminal penalties and clarifying moral rights infringement boundaries in AI learning, parody, and modified uses. He added, “Since it is technically difficult to restrict the use of copyrighted works solely for AI learning purposes, a system that establishes and provides reliable public learning data could serve as an interim solution.”

Appreciation Plaque Ceremony

and General Discussion

“Balanced Copyright Protection is

Important in the Era of AI”

Co-chairs Kim Byungil and Lee Sungyeop received plaques of appreciation for their leadership in this year’s forum, while Professor Lee Kyuho was appointed as the next chair. The general discussion broadly covered key issues in copyright protection in the era of AI.

Professor Kim Incheol of Sangmyung University raised the question of how to interpret cases treating generative AI as a “tool,” pointing to the German GEMA case to highlight the need for discussion on the future direction of moral rights litigation structure (voluntary litigation) in the AI era. Jung Hyunsoon, a judicial researcher at the Supreme Court, noted that web crawling for AI learning often mixes legal and illegal content, questioning whether the U.S. case law, which denies fair use in learning from pirated copies, applies directly to this situation.

Professor Kim Byungpil of KAIST proposed permitting limited fair use in large-scale model training, coupled with service-level technological protection measures, such as filtering and sorting, to prevent the reproduction of copyrighted works, alongside a structure holding users accountable for entering malicious prompts. Yoo Byunghan, president of the Korea Software Property Right Council, stated that the growing sophistication of prompt engineering will heighten conflicts between copyright and industrial property rights. He emphasized the need for a mid- to long-term legal and institutional roadmap that takes into account the era of AGI.

Co-chair Kim wrapped up the discussion by stating, “In an era where coexistence with AI is inevitable, the key task is to find a balance among copyright, moral rights, fair use, and industrial development.”