Taiwan AI Corpus Growth Stalled by Copyright Payment Disputes

Taiwan’s national AI training corpus expansion has stalled as government agencies and private rights holders remain deadlocked over compensation terms for copyrighted content. The National Science and Technology Council confirmed that while public domain data integration continues, acquiring high-quality private datasets is delayed due to unresolved payment disputes with publishers and individual creators. Industry stakeholders allege current government valuation models undervalue intellectual property essential for competitive large language model development. Officials maintain that budget constraints and ambiguous legal standards for AI training fair use complicate immediate settlements. This impasse threatens the timeline for Taiwan’s domestic foundation model roadmap, which relies on diverse local linguistic data to reduce dependence on foreign AI infrastructure. Negotiations are ongoing, but no revised licensing framework has been finalized. The situation highlights the growing friction between national AI sovereignty goals and existing intellectual property regimes across Asia.