The Pulse
South Korea Routes Government-Funded AI Data Through AI Hub
South Korea’s Cabinet approved amendments requiring government-funded AI training data to be linked to the AI Hub, with exceptions for protected information. A separate amendment would strengthen the One-Window system for finding data held
AI.info Team · Written with AI ·
“To strengthen AI competitiveness, it is important to secure high-quality AI training data and make it easier for people to find and use the data they need.”
Um Yeol, director of the Ministry of Science and ICT’s AI Policy Office
Two amendments target data access
South Korea’s Cabinet approved amendments to the AI Basic Act and the Data Industry Act on October 13, setting out new rules for government-supported AI training data and a stronger system for locating public and private datasets. The Ministry of Science and ICT says the changes are intended to make publicly funded training resources easier to find and use while improving connections among data platforms.
The AI Basic Act amendment would require AI training data built or generated through government projects to be linked to the AI Hub, the national platform for AI data. The requirement concerns data produced during project work; it does not mean every dataset becomes openly available without conditions. The amendments are bills approved by the Cabinet, not rules already in force.
That distinction matters for researchers and companies planning to use the material. The data-sharing provision concerns access to training datasets, while the second amendment focuses on information about where data is held. One-Window, the government’s platform for locating data across participating systems, is meant to help users search across those holdings rather than replace each platform’s underlying collection.
945 government-funded datasets are already open
The AI Hub has operated since 2018, and 945 AI training datasets built through government projects were open as of the end of 2025, according to reporting on the ministry’s announcement. The amendment would establish a legal basis for connecting and providing government-project datasets through the platform, alongside measures for administrative, financial and technical support and quality management.
The proposal also sets out exceptions. Data may be withheld when project participants determine that sharing would risk trade secrets, violate privacy or copyright protections, or affect national security, among other grounds to be specified by presidential decree. Those safeguards mean the new duty is not an unconditional disclosure rule.
One-Window’s reach is smaller than the total number of datasets in AI Hub: the government reported 88 connected data platforms as of August 2026, comprising 73 public platforms and 15 private ones. The Data Industry Act amendment would establish a legal basis for the integrated platform and allow the ministry to request cooperation on connecting information about where data is held. Public institutions would generally have to comply unless they have a specific reason not to; the amendment also provides for technical and financial support.
Implementation is scheduled for April 2027
Both amendments are scheduled to take effect six months after promulgation, in April 2027. Until then, their provisions do not create an active duty for project operators to connect data to AI Hub or for institutions to respond to One-Window cooperation requests. The Ministry of Science and ICT is expected to prepare subordinate rules ahead of implementation.
The practical test will be whether the new duties make usable datasets easier to discover without erasing the protections written into the bills. The government’s existing figures offer a starting point: 945 government-funded training datasets open on AI Hub at the end of 2025, and 88 platforms connected to One-Window in August 2026. The amendments set a framework for expanding those connections, while leaving specific exclusions and implementation details to follow-on rules.