Proposal

APrIGF 2026 Session Proposal Submission Form
Part 1 - Lead Organizer
Contact Person
Ms. Sodam KUM
Email
Organization / Affiliation (Please state "Individual" if appropriate) *
Ewha Womans Univeristy / KISA(Korea Internet & Security Agency) EG@IG
Designation
Student
Gender
Female
Economy of Residence
South Korea
Stakeholder Group
Youth
List Your Organizing Partners (if any)
KISA (Korea Internet & Security Agency) EG@IG (Expert Group at Internet Governance)
Part 2 - Session Proposal
Session Title
Governing AI Training Data in the Asia-Pacific: Transparency, Verifiability, and Creator Rights
Thematic Track of Your Session

  • Option

    • Primary: AI Governance for Good
    • Secondary: Digital Resilience and Building of Trust
Session Format
Showcase (30 minutes)
Where do you plan to organize your session?
Hybrid (with onsite moderator and speakers both online and offline, with online moderator for questions and comments from remote participants)
Specific Issues for Discussion
This session will explore the transparency and verifiability of AI training data. First, it will examine national AI strategies in Asia-Pacific countries. Second, in the context of the Books3 dataset controversy, it will discuss the challenges authors face in determining whether their works have been included in AI training datasets. Third, it will discuss Japan’s Copyright Act Article, which has become an important reference point in debates on text and data mining exceptions. Fourth, it will examine Korea’s AI Basic Act, which introduces transparency obligations for certain AI systems but does not provide a comprehensive framework for training data disclosure and opt-out rights. Beyond national strategies, the session will examine the limits of the existing international copyright framework itself. The Berne Convention (1886), the WIPO Copyright Treaty (1996), and the WIPO Performances and Phonograms Treaty (1996) were designed to secure automatic, cross-border protection and to extend authors' rights into the online environment. Additionally, the EU has constructed its own AI Act for the sustainable and ethical use of AI. However, none of these instruments anticipated the use of generative AI for training. There is no shared norm addressing the transparency, verifiability, or opt-out of AI training data. This gap is especially pronounced for countries such as Nepal and other developing countries in the Asia-Pacific region. The session will also address the limits of technical solutions such as membership inference attacks, watermarking, data provenance tools, and dataset documentation. Still, they do not yet provide a reliable and accessible mechanism for users to verify whether specific work was used in model training. Therefore, the issue should be understood not only as a technical problem but also as a governance problem concerning who controls training data, who has the authority to verify it, and how accountability should be distributed.
Describe the Relevance of Your Session to APrIGF
This showcase advances APrIGF's ‘Safeguarding a Resilient Internet in Times of Crisis’ theme through training data transparency, accountability, human oversight, and data governance. As AI becomes embedded in education, culture, media, and knowledge production, training data governance is now a foundation for public trust. Many APAC economies no longer merely adopt AI developed elsewhere; they actively build sovereign AI capabilities, national foundation models, and multilingual datasets. Yet training data routinely crosses borders, raising unresolved questions about copyright, consent, cultural representation, and creators' rights when works are used without notice or verification. Compared with the EU or transatlantic spaces, the region has fewer mechanisms to coordinate this governance, and spans diverse legal and copyright traditions: Korea, Japan, the EU, and Singapore show divergent rules, while India, host Nepal, and others still lack AI-specific data governance. By comparing regional examples, the showcase moves beyond a narrow copyright dispute and asks how transparency and accountability can be implemented in practice. It weighs two pathways: national legislative reform, and a shared normative layer building on the cross-border principles of the Berne Convention (1886), the WIPO Copyright Treaty (1996), and the WIPO Performances and Phonograms Treaty (1996). It asks how governments, companies, technical experts, civil society, creators, and youth can co-design principles that are innovation-friendly and respect rights. The outcome is practical regional principles: transparency, verifiability, creator rights, interoperable opt-out, independent audit, and multi-stakeholder oversight. Ultimately, it supports APrIGF's goal of an open, inclusive, resilient, and trustworthy Internet, framing training data governance as strategic for AI sovereignty, digital trust, and public-interest technology in the Asia-Pacific.
Methodology / Agenda (Please add rows by clicking "+" on the right)
Time frame (e.g. 5 minutes, 20 minutes, should add up to the time limit of your selected session format) Description
5 minutes (Opening and framing) The moderator introduces the session's objective and its relevance to APrIGF, then frames the central question: how can the Asia-Pacific reconcile sovereign AI ambitions with accountable, cross-border training data governance? Speakers are briefly introduced.
15 minutes (Issue presentation) The lead presenter gives a short talk on AI training data transparency and verifiability, explaining the black-box dataset problem. It maps the regional landscape—Japan, Korea, Nepal—and shows why national law and technical tools alone fall short.
5 minutes (Audience participation) The moderator opens the floor for questions and comments, and may run a quick poll on which governance principles the audience would prioritize for the APAC region. Speakers respond from their respective stakeholder perspectives.
5 minutes (Wrap-up and expected outcomes) The moderator summarizes the key points and identifies possible regional principles for AI training data governance—transparency, verifiability, creator rights, opt-out, and multi-stakeholder oversight.
Moderators & Speakers Info (Please complete where possible) - (Required)

  • Moderator (Primary)

    • Name: Sodam KUM
    • Organization: KISA EG@IG
    • Designation: Student
    • Gender: cis woman
    • Economy / Country of Residence: Republic of Korea
    • Stakeholder Group: Youth / Students
    • Expected Presence: In-person
    • Status of Confirmation: Confirmed

  • Moderator (Facilitator)

    • Name: Seungwoo Jang
    • Organization: KISA EG@IG
    • Designation: Student
    • Gender: cis man
    • Economy / Country of Residence: Republic of Korea
    • Stakeholder Group: Youth / Students
    • Expected Presence: In-person
    • Status of Confirmation: Confirmed

  • Speaker 1

    • Name: Seoyeon HAN
    • Organization: KISA EG@IG
    • Designation: Student
    • Gender: cis woman
    • Economy / Country of Residence: Republic of Korea
    • Stakeholder Group: Youth / Students
    • Expected Presence: In-person
    • Status of Confirmation: Confirmed

  • Speaker 2

    • Name: Eunsun Lee
    • Organization: KISA EG@IG
    • Designation: Student
    • Gender: cis woman
    • Economy / Country of Residence: Republic Korea
    • Stakeholder Group: Youth / Students
    • Expected Presence: In-person
    • Status of Confirmation: Confirmed

  • Speaker 3

    • Name: Sehyeon Kim
    • Organization: KISA EG@IG
    • Designation: Student
    • Gender: cis man
    • Economy / Country of Residence: Republic of Korea
    • Stakeholder Group: Youth / Students
    • Expected Presence: In-person
    • Status of Confirmation: Confirmed

  • Speaker 4

    • Stakeholder Group: Select One
    • Expected Presence: Select One
    • Status of Confirmation: Select One

  • Speaker 5

    • Stakeholder Group: Select One
    • Expected Presence: Select One
    • Status of Confirmation: Select One
Please explain the rationale for choosing each of the above contributors to the session.
The contributors to this session are members of the KISA EG@IG Intellectual Property Rights and AI Team. The team has been conducting research and discussion on the intersection of AI training data, copyright, transparency, and governance, with a particular focus on how creators and rights-holders can verify whether their works have been used in AI training datasets. This proposal builds on the team’s previous session at KrIGF, where the discussion focused on limitations of Korea’s current legal framework. Through the KrIGF session, the team identified that the issue is not only a domestic copyright question, but also a broader Internet governance issue involving data control, platform power, verification authority, public trust, and multi-stakeholder accountability. For APrIGF, the team expands this prior work from the Korean context to the Asia-Pacific regional context.
Has AI been used to develop this proposal?
Yes
How:
to adjust the max character of the proposal and correct the grammar
Please declare if you have any potential conflict of interest with the Program Committee 2026.
No
Are you or other session contributors planning to apply for the APrIGF Fellowship Program 2026?
Yes
Upon evaluation by the Program Committee, your session proposal may only be selected under the condition that you will accept the suggestion of merging with another proposal with similar topics. Please state your preference below:
Yes, I am willing to work with another session proposer on a suggested merger.
Number of Attendees (Please fill in numbers)
    Gender Balance in Moderators/Speakers (Please fill in numbers)
      Consent
      I agree that my data can be submitted to forms.for.asia and processed by APrIGF organizers for the program selection of APrIGF 2026.