To Their Credit

South Korea puts AI transparency and training-data rights at copyright conference

The 16th ICOTEC meeting will examine watermarks, AI development records and tools intended to curb unauthorised model training.

Ledger: DisputedNo outcome yet

South Korea’s Ministry of Culture, Sports and Tourism will hold the 2026 International Copyright Technology Conference in Seoul on Nov. 4, putting traceability of AI-generated material and safeguards for copyrighted training data at the centre of its copyright-technology meeting.

The 16th International Copyright Technology Conference, known as ICOTEC, will be held at JW Marriott Dongdaemun Square Seoul in collaboration with the Korea Copyright Commission and Korea Copyright Protection Agency. Launched in 2011, the event introduces copyright technologies from Korea and abroad and discusses related issues.

The programme will be organised around “Transparency in Generative AI” and “Protecting Rights in Training Data and Copyright Technology.” Asiae reported that the conference is scheduled as part of November Copyright Week, the ministry’s newly integrated copyright-awareness campaign, which builds on Copyright Festival Month, held since 2024.

Hanyang University’s Lee Sang-wook will deliver the keynote, “Rethinking Creativity and Copyright in the Age of Generative AI.” Garrett Levin, a senior intellectual-property legal and policy officer at the World Intellectual Property Organization, will address copyright’s foundations and international technological cooperation in the generative-AI era.

The transparency session will examine standards intended to improve the reliability of AI-generated content and their possible industry application. Korea University’s Kim Seong-ung will introduce watermarking that embeds identifying information in AI outputs, while Adobe senior director Andy Parsons will address methods for establishing the reliability of digital content.

SK Telecom manager Jang Hak-seong will present an AI Bill of Materials, or AI-BOM, as a transparency mechanism. An AI-BOM records the data, models, software and other resources used in AI development, potentially creating a more legible account of how a system was built.

A separate session will consider technical responses to unauthorized AI training and copyright disputes. Newtoon chief executive Lee Jong-pil will discuss identifying the sources of training data and creating a licensing framework. Sangmyung University’s Kang Sang-wook will present machine unlearning, which removes particular training information from a model, alongside knowledge-distillation technology; Kim Young-sik of the Daegu Gyeongbuk Institute of Science and Technology will address adversarial noise and adaptive watermarking.

The meeting follows the ministry’s publication of an English-language fair-use guide for generative-AI training in May. The official guide says commercial training and training involving web crawling are not automatically excluded from fair use; courts must assess particular cases by weighing four factors, and the guide’s examples are not authoritative legal interpretations.

The ministry will also recognise contributors to copyright protection and technology development. Hanyang University’s Jung Hae-jun and Software Policy & Research Institute principal researcher Kwon Young-hwan will receive Culture Ministry awards, while LG AI Research will receive the Chairperson of the Presidential Council on Intellectual Property Award and Korea University’s Kim Sang-pil will receive the WIPO Award.

Choi Young-jin, the ministry’s director general for copyright policy, said copyright technology was increasingly important both for protecting rights and for creating fair conditions for the use of training data. The ministry said it would take account of technological developments and views discussed at the conference in future copyright-technology policy.

See also

More news

All news
  1. German study sees revenue slump and job losses if platforms control news accessLedger: Disputed
  2. BeatMelo redesigns its AI music and video studioLedger: Adopted
  3. AI news use triples as publishers adapt to answer enginesLedger: Adopted