On 25 April 2025, the European Economic and Trade Office (EETO) and Taiwan Intellectual Property Office (TIPO) co-organized a seminar in Taipei on Generative AI and Copyright, gathering experts from both the EU and Taiwan across public and private sectors.
The opening remarks set the stage for a thoughtful and forward-looking discussion. Lutz Guellner, Head of the EETO, described AI as both the most used and least understood term: one that reflects society’s hopes, fears, and ambitions. He emphasized the EU’s commitment to safety, human dignity, and fair systems, especially in the context of the new AI Act. Cheng-Wei Liao, Director General of TIPO, called for greater transparency around training data, stronger opt-out rights for copyright holders, and pointed to the value of learning from the EU’s evolving regulatory experience.
The first session focused on legal frameworks and copyright issues surrounding Generative AI. Anneli Andresson-Bourgey of the European Commission explained how the EU’s Copyright Directive and AI Act function in tandem to ensure respect for creators while supporting innovation. She highlighted the opt-out mechanism for text and data mining and obligations for developers to disclose training data. Chia-Hung Kao from TIPO shared Taiwan’s perspective, explaining the legal distinctions between AI’s training and output stages, and reinforcing that only human-generated works are eligible for copyright. He also discussed emerging international practices, including licensing agreements with news outlets and image libraries, and Taiwan’s plans for a sovereign AI training dataset.
A short networking break followed, where attendees were invited to enjoy refreshments, provided space for participants to exchange views and continue the conversation in a more relaxed setting.
The second session turned toward the practical challenges of developing and deploying Generative AI. Dr. Amit Datta of Aleph Alpha shared insights from the European AI industry, outlining key regulatory obligations under the EU’s AI Act, GDPR, and copyright rules. He emphasized the importance of access to high-quality data, infrastructure challenges, and the need for industry-wide Codes of Practice. Dr. Chin-Yuan Fan of NARLabs introduced Taiwan’s TAIDE project, a national effort to develop culturally and linguistically aligned AI systems. He detailed data sourcing efforts, engagement with content owners, and the project’s broader goal of building trustworthy, locally rooted models.
The seminar concluded with a shared recognition of the importance of cross-border collaboration in shaping responsible AI development. With both the EU and Taiwan actively addressing the legal, ethical, and technical dimensions of Generative AI, the event served as a timely platform for mutual learning and continued partnership.
