🐯Stalecollected in 27m

Korea India copy Chinese AI in global race

Korea India copy Chinese AI in global race
PostLinkedIn
🐯Read original on 虎嗅
#ai-copying#geopolitics#national-aiqwen-/-deepseek-/-zhipu-aiqwendeepseekzhipu-aiupstagenaver

💡Korea $2.5B AI bid fails over Chinese model code copies—open-source pitfalls.

⚡ 30-Second TL;DR

What Changed

Korean gov's 5300B KRW contest: 3 teams copied Qwen, DeepSeek, Zhipu code.

Why It Matters

Exposes reliance on Chinese open-source LLMs by rivals; questions national AI sovereignty amid funding/talent gaps. Boosts visibility of Qwen/DeepSeek globally.

What To Do Next

Audit your LLM codebase for Zhipu/DeepSeek licenses before national grants.

Who should care:Researchers & Academics

Key Points

  • Korean gov's 5300B KRW contest: 3 teams copied Qwen, DeepSeek, Zhipu code.
  • Upstage admitted using Zhipu open components; NAVER used Alibaba encoders.
  • Japan 6/10 top models based on DeepSeek/Qwen; India faked Unitree Go2 as own.
  • Challenges: funding shortages, GPU scarcity, Korean data lack.
  • Gov continues despite scandal, adds revival round.

🧠 Deep Insight

Background and context from public sources — not the original article. 7 sources cited.

🔑 Enhanced Key Takeaways

  • South Korea's government-funded AI competition explicitly lacks rules prohibiting foreign open-source code use, creating ambiguity that enabled finalists to incorporate components from Chinese models like Zhipu AI without violating stated guidelines[1][3].
  • The controversy triggered a broader debate within South Korea's AI ecosystem about the feasibility of building sovereign AI from scratch versus pragmatically leveraging existing open-source technologies, with government officials framing the technical dispute as evidence of healthy industry self-regulation[5].
  • Beyond Upstage's Zhipu integration, Naver acknowledged using external encoders (attributed to Alibaba in the article summary) and SK Telecom made comparable admissions, indicating systemic reliance on foreign foundational technologies across multiple finalists[1].

🔮 Future ImplicationsAI analysis grounded in cited sources

South Korea's sovereign AI initiative will require explicit policy clarification on open-source code permissibility to prevent future accusations and ensure fair competition.
The Ministry of Science and Technology has not issued new guidance since the January 2026 controversy despite intense debate, leaving the competition's legitimacy vulnerable to repeated challenges[1][3].
The reliance on foreign AI components by multiple finalists suggests that achieving genuine technological independence within the stated timeline (winners selected by 2027) may require either extended deadlines or redefined success metrics.
Three of five finalists used foreign code, and companies argue that building entirely from scratch is impractical, indicating structural barriers to the original competition objectives[1].

Timeline

2025-06
South Korean government launches three-year 'AI national team' competition to develop independent domestic AI models
2026-01-01
Sionic AI CEO Ko Suk-hyun publicly accuses Upstage of using Zhipu AI open-source code with retained copyright markers in its Solar Open 100B model
2026-01-03
Ko Suk-hyun issues public apology after Upstage conducts live verification demonstrating independent model development; government officials praise the technical debate as evidence of ecosystem health
📰

Weekly AI Recap

Read this week's curated digest of top AI events →

👉Related Updates

AI-curated news aggregator. All content rights belong to original publishers.
Original source: 虎嗅

This is a summary, not the original. Read the source, or get the weekly briefing.

Weekly AI briefing

One email a week. Unsubscribe anytime.