Reke Wang / Secretary General of Wikimedia Taiwan
On May 20, just three days after the end of ESEAP Conference 2026 — and only the second day after returning from Kaohsiung to Taipei — I participated, in my capacity as Secretary-General of the Wikimedia Taiwan, in the “Web Crawling Governance Policy Dialogue” organized by the Institute for Information Industry (III). The meeting was convened in response to the rapid growth of generative AI and the increasing need for clearer authorization and cooperation mechanisms between AI crawlers collecting training data and content producers. The Taiwanese government hopes to gather perspectives from different sectors as references for future policy research. Even though I had only just begun recovering physically from the conference, it was still important to attend on behalf of the association.
The meeting was conducted through group discussions, and I was assigned to the group focused on public-interest databases and public-interest platforms. Participants in the group included representatives from collaborative fact-checking communities, open data companies, government public databases, cybersecurity service providers, and legal professionals.
During the discussion, I shared data and policy approaches published by the Wikimedia Foundation regarding AI crawlers. I explained how Wikimedia projects attract AI companies toward more suitable structured datasets for AI training, thereby helping protect limited system resources. Through the exchanges among participants, the discussion gradually converged on the idea that even open or public-interest databases — despite welcoming the use of their content — still require sustainable revenue-sharing mechanisms to secure additional resources.
I also pointed out that Wikipedia has increasingly been regarded as an important platform for AEO (Answer Engine Optimization). While we do not want Wikipedia to become a promotional platform, this trend also demonstrates that although AI may replace some of Wikipedia’s direct traffic, it simultaneously expands Wikimedia’s influence. This situation differs significantly from the concerns faced by commercial media organizations whose primary issue is direct profit loss, and such complexities should be carefully considered in policymaking.
The group also examined possible legal tools and found that criminal-law approaches may be difficult to enforce against cross-border AI services. Civil mechanisms, such as terms-of-use agreements, appear more practical. However, most non-profit databases lack the resources that Wikimedia possesses to invest in AI-oriented commercial datasets. As a result, there may be a need for collective licensing structures similar to copyright collective management organizations, with intermediary licensing bodies helping manage authorization on behalf of smaller organizations. If similar organizations existed internationally and established cooperative agreements with one another, protections could become more comprehensive.
Although Wikimedia itself may not need to participate in such a licensing organization, the Wikimedia Foundation’s cooperation models and licensing arrangements with AI companies could become valuable references for future frameworks.
The Wikimedia Taiwan will continue to assert itself as a major stakeholder in AI policy discussions and actively participate in dialogue across sectors, helping ensure that future AI policies support the development of open knowledge in the new era.
Can you help us translate this article?
In order for this article to reach as many people as possible we would like your help. Can you translate this article to get the message out?