SEOUL—The Korean Association of Newspapers (KAN), led by President Lim Chae-cheong, has intensified its push for the revision of the nation’s foundational AI legislation, calling for a mandatory provision that would force Artificial Intelligence companies to disclose their training data. The move is a significant step in the ongoing global debate over copyright protection and transparency in the rapidly expanding generative AI market.

Call for Transparency in AI Framework Act 

On Wednesday, the KAN submitted a formal opinion letter to key government bodies, including the National Assembly’s Science, ICT, Broadcasting and Communications Committee, the Ministry of Science and ICT, and the Ministry of Culture, Sports and Tourism. The association urged authorities to amend the upcoming Framework Act on the Development and Establishment of Trust in Artificial Intelligence (commonly referred to as the AI Basic Act), which is set to take effect in January 2026.

The KAN argues that while the government is currently preparing subsidiary legislation for the Act, the absence of a clear legal mandate for the disclosure of training data poses a critical limitation to copyright protection and ensuring the transparency of AI models. This lack of explicit regulation is seen as a major vulnerability, particularly for media organizations whose copyrighted content is frequently used to train large language models (LLMs) without authorization or clear compensation.

The association, along with four other media groups, previously voiced this concern in December last year when the AI Basic Act was passed by the National Assembly. At the time, legislators agreed to prioritize the bill's passage but acknowledged the need to address deficiencies, such as data transparency, through subsequent amendments.

Aligning with Global Standards: The EU AI Act 

The KAN strongly emphasized the need for South Korea to align its regulations with international standards, citing the European Union’s approach. The EU's groundbreaking Artificial Intelligence Act (AI Act), which entered into force in August, requires developers of General-Purpose AI (GPAI) models to publish a sufficiently detailed summary about the content used for training, especially copyrighted materials.

"To protect copyright, secure the transparency of AI technology, and meet global standards, Korea must stipulate a similar provision for training data disclosure in its AI Basic Act," the KAN stated in its opinion.

Despite the consensus among lawmakers to address the issue, the KAN points out that discussions for a revision have been sluggish, even as the regulatory framework is being finalized. The industry views mandatory data transparency as a crucial mechanism to allow copyright holders to verify the usage of their intellectual property, thereby fostering a more responsible and trustworthy AI ecosystem.

The KAN’s demand highlights the growing tension between promoting rapid AI industrial development and safeguarding the rights of content creators. As Korea positions itself as a global leader in AI innovation, observers anticipate that legislative action on mandatory data transparency will be the next major hurdle in the country's AI governance journey.