South Korea to release 25 high-value AI public datasets this year, targeting 100 by 2028

NOL, the accommodation booking platform formerly known as Yanolja, uses fire facility data released by the National Fire Agency to inform users whether lodging properties have sprinklers and other fire suppression systems installed.

LBOX, a legal AI solutions company, draws on statutory interpretation cases and special administrative adjudication rulings published by the Ministry of Government Legislation to train its legal AI and generate response references.

The Ministry of Interior and Safety said Monday it has finalized detailed project plans for 25 datasets to be released this year as part of its "AI and High-Value Public Data TOP 100" initiative, and that the rollout is now underway.

The datasets were chosen from roughly 3,280 candidate projects identified through on-site surveys of more than 800 companies last year and a nationwide online demand survey. Final selections were made by an external panel of experts who weighed economic impact, alignment with national policy priorities and AI compatibility.

The ministry released 10 datasets last year and plans to add 25 this year, 30 in 2027 and 35 in 2028, bringing the cumulative total to roughly 100 high-value datasets by 2028.

This year's releases are organized around four core sectors directly tied to new private-sector business model creation.

In the new industries sector, the Korea Institute of Energy Research will open its renewable energy technical potential dataset, which maps the generation potential of seven energy types — solar, wind, hydro, marine, biomass, waste and geothermal — by latitude, longitude and administrative district.

In the K-culture sector, the Korea Culture Information Service will provide AI training data covering traditional architecture, including 3D models and 2D images of dancheong decorative patterns on historic buildings, along with period codes, symbolic meanings and supplementary descriptions of cultural artifacts for AI learning.

In the disaster and safety sector, the Korea Infrastructure Safety Corporation will share inspection data on special-type bridges such as suspension and cable-stayed bridges. The data includes damage imagery, damage type, cause and repair recommendations, as well as detection data on vehicles crossing those bridges, including timestamps and locations.

In the AI training sector, the Korea Fair Trade Commission will release AI training data optimized from its rulings on violations of the Fair Trade Act and related laws, while the Rural Development Administration will provide crop disease and pest diagnostic data combining images of disease symptoms and adult and larval forms of pests with detailed descriptive annotations.

Notably, datasets containing personal information — including family and youth counseling call records and transport worker licensing data — will be released as synthetic data, described as simulated datasets that replicate the characteristics, structure and distribution of the original, to ensure both privacy protection and practical usability.

"We will swiftly open the data AI companies need to support their growth and help realize South Korea's goal of becoming one of the world's top three AI powers," Minister of Interior and Safety Yun Ho-jung said. "We will strengthen the release of training data through additional demand surveys and transition to an AI-friendly public data management framework to pursue public data policies that keep pace with the times."