[Edaily Reporter JAEMIN SONG ] LG Corp. AI Research has developed a technology that reduces the cost of analyzing copyright and personal information infringement risks in artificial intelligence (AI) training data to one-thousandth of the cost of existing methods. The technology, which automatically analyzes legal risks such as copyright and personal information contained in AI, received the Chairman’s Award from the National Intellectual Property Commission. As copyright disputes and personal information protection issues surrounding AI training data continue to spread, this technology is expected to significantly reduce the time and cost required to assess legal risks.

LG Corp. announced on the 4th that it received the Chairman’s Award from the Korea Intellectual Property Office at the “2026 International Copyright Technology Conference (ICOTEC)” Copyright Technology Awards, hosted by the Ministry of Culture, Sports and Tourism, in recognition of the technical capabilities of its self-developed Agent AI compliance agent, “ExaOne Nexus.”

ExaOne Nexus is an AI agent that automatically analyzes risks related to copyright, licensing, and personal information contained in large-scale AI training datasets. It tracks the origin of datasets to identify changes in rights relationships that occur when original data is redistributed or combined with other data. ExaOne Nexus was developed in response to the recent emergence of legal risks stemming from controversies over copyright infringement in AI training data.

Its most notable feature is that it has dramatically reduced analysis costs. According to LG Corp. AI Research, the cost of analyzing a single dataset using ExaOne Nexus is approximately one-thousandth that of conventional methods. It also analyzes data 45 times faster than legal experts and achieved 26% higher accuracy. This solution simultaneously reduces both the cost and time required to review the legal risks associated with the vast amounts of training data used in AI model development.

ExaOne Nexus evaluates the legal safety of data based on 18 criteria established using actual legal disputes and case law. Each criterion is scored on a scale from 1 (high risk) to 5 (low risk), and after a comprehensive evaluation, the dataset is classified into seven levels ranging from “very safe” to “very risky.”

LG Corp. AI Research has also established a compliance framework that reflects regulations from major countries, including South Korea’s Framework Act on Artificial Intelligence and the European Union’s (EU) AI Act. In collaboration with approximately 40 attorneys from major domestic and international law firms, the legal validity of the analysis results was verified, and the risk assessment results for over 20,000 open datasets were made publicly available to the academic and research communities.

Currently, ExaOne Nexus is applied as a mandatory verification process for all AI models developed by LG Corp. AI Research. It identifies copyright and personal information risks throughout the entire AI development process—from data collection and refinement to model development and service deployment. Going forward, the company plans to expand its supply to overseas markets, including global Big Tech companies and law firms.

In addition, LG Corp. AI Research has secured or filed 32 related patents domestically and internationally, covering areas such as the removal of illegal training data and the detection of copyright infringement in AI-generated content. Since its establishment in 2020, the institute has filed a total of 1,131 patents, including 537 in Korea and 594 overseas.

Lee Hwa-young, Head of the AI Business Development Division at LG Corp. AI Research, stated, “Amid the rapid advancement of AI technology, securing social trust is a core competitive advantage for companies,” adding, “We will continue to introduce AI technologies that combine high reliability with outstanding performance.”