LG AI Research's EXAONE Nexus Cuts Copyright Vetting Cost 1,000-Fold

What EXAONE Nexus does

EXAONE Nexus is an AI agent that analyzes copyright, licensing and personal-information risks in large-scale training datasets. According to LG AI Research, it works 45 times faster than legal experts when analyzing a single dataset, with 26% higher accuracy and a cost of about 1/1,000.

The institute built its own compliance framework reflecting major regulations, including South Korea's AI Basic Act and the EU AI Act, a design aimed at global companies facing different rules in each country. To back up the results, it worked with about 40 lawyers at large law firms in the United States and elsewhere to check the legal validity of the analysis. Risk assessments for roughly 20,000 open datasets used in validation have been released to the academic and research community.

Internal use first, then export

LG AI Research has made EXAONE Nexus a mandatory check for every AI model it develops internally, covering data collection and cleaning, model development and service deployment. With internal adoption complete, it plans to extend supply to global big tech firms and law firms. It has not disclosed a timeline or the size of the business.

Disputes over copyright in AI training data have become a major legal risk for companies, and reviewing huge datasets by hand is impractical, which ties verification directly to the cost of building AI.

Lee Hwa-young, head of AI business development at the institute, said earning social trust is a core competitive strength for companies amid rapid AI progress, and promised to keep delivering reliable, high-performing AI.

Since its founding in 2020, the institute has filed 1,131 patents, 537 in South Korea and 594 abroad. It has also joined more than 30 global discussions on AI ethics and governance.