Several AI Models in South Korea Face Allegations of Resembling Chinese Models, with Suspected Code Plagiarism
2026-01-16 / Read about 0 minute
Author:小编   

In June 2025, the South Korean government initiated a competition centered around artificial intelligence foundational models. The goal was to identify two domestic AI models by 2027 that are entirely rooted in South Korean technology and capable of rivaling the performance of leading international models. However, the competition has recently become mired in controversy regarding 'technological autonomy.' Among the five finalist companies, three have been outed for incorporating foreign open-source components, with several models displaying a striking resemblance to Chinese open-source models.

Specifically, Upstage has come under fire for having components that bear a strong resemblance to those found in China's Zhipu AI open-source model. Shockingly, Zhipu AI's copyright marks were still discernible within the code. Naver's visual and audio encoders have also raised eyebrows, with allegations suggesting they bear a resemblance to products from Alibaba and OpenAI. Meanwhile, SK Telecom's inference code structure has drawn suspicion for its similarity to that of DeepSeek.

Despite these companies vehemently denying any form of 'shell development' and underscoring their independent development of core modules, the controversy shows no signs of abating. The South Korean government, undeterred, is forging ahead with the competition. It has called for a fair and transparent review process to ensure that the project's objectives remain firmly aligned with the original vision of 'autonomous AI.'