KIPRIS at 30: Korea Tests AI Natural-Language Patent Search with a 15-Member Panel

Korea's Ministry of Intellectual Property (the former KIPO) and the Korea Institute of Patent Information marked the 30th anniversary of KIPRIS at an event in Seoul on 22 September. KIPRIS is Korea's public IP information search service, holding about 150 million IP records collected by the ministry from Korea and 29 countries and open to anyone free of charge. At the event's talk session with young users, 6 researchers, founders, students and patent attorneys drew on their own experience to propose an AI natural-language search that lets beginners find patents in everyday language without professional search formulas, and a function that summarises and explains complex patent documents at the user's level; they also called for more hands-on IP training and wider opening of IP data to the private sector. Minister Kim Yong-seon said the ministry would reflect the young users' views in the service, actively introduce AI natural-language search and expand overseas IP information.
One distinction matters: the AI natural-language search is not yet generally available. The KIPRIS national evaluation panel appointed the same day, 15 patent attorneys, researchers and students, will try out an AI-based search service that is still under development and give feedback from the user's perspective; the service is a project selected under the Ministry of Science and ICT's public AX programme and is being developed by the ministry with private-sector partners. What the ministry has confirmed is a direction and a testing arrangement, not a launch. J&C recommends that companies with Korean patent filings or a need to monitor Korean competitors keep relying on KIPRIS's existing search-formula and classification searches for now and not treat results from the AI natural-language search, still in development and testing, as evidence of search completeness; at the same time, watch the data interfaces the Ministry of Intellectual Property opens next and assess whether KIPRIS data can be fed into their own monitoring workflows.



