Blog
tech
7 min read

"News Isn't Free": The Beginning of KBS·MBC·SBS's Copyright War Against OpenAI

On February 23, 2026, Korea's three major terrestrial broadcasters filed a copyright infringement lawsuit against OpenAI at the Seoul Central District Court. This is the first legal action by Korean broadcasters against a global AI company and a test case for how Korean courts will address the copyright gap in AI training data that the Artificial Intelligence Basic Act has not filled.

Decades of News Content Used for AI Training Without Permission

On February 23, 2026, the AI copyright war that had been raging in American and European courts moved to the Seoul Central District Court. Korea's three major terrestrial broadcasters—KBS, MBC, and SBS—filed a lawsuit against OpenAI seeking injunctive relief and damages for copyright infringement.

The three broadcasters claim that OpenAI's generative AI service ChatGPT used their news content without authorization for training purposes. According to the Korea Broadcasting Association, this is the first copyright lawsuit filed by domestic broadcasters against a global AI company.

TL;DR (3-Line Summary)

  • Lawsuit Background: Korea's three terrestrial broadcasters take legal action against OpenAI's unauthorized news training—first lawsuit by domestic broadcasters against global AI company
  • Core Issue: OpenAI has signed paid licensing agreements with global media outlets while refusing to negotiate with Korean broadcasters
  • Future Outlook: Overseas precedents like Anthropic's $1.5 billion settlement and Germany's GEMA victory may influence Korean court decisions

Facts: What Happened

Timing and Parties of the Lawsuit

On February 23, 2026, KBS, SBS, and MBC filed a lawsuit against OpenAI at the Seoul Central District Court seeking injunctive relief and damages for copyright infringement. The Korea Broadcasting Association is coordinating the litigation, with the three networks forming a united front.

Alleged Infringement

The three broadcasters claim that OpenAI used news content produced by their networks without permission during ChatGPT training. Specifically:

  • Decades of accumulated broadcast news archives were used as training data
  • Content was accessed by circumventing paywalls and copyright notices
  • Used for commercial AI services without compensation

OpenAI's Double Standard

According to Korea Herald reports, the three broadcasters claim that OpenAI has signed paid licensing agreements with global media outlets while refusing to negotiate with Korean broadcasters. In fact, OpenAI has:

  • Signed paid licensing agreements with Associated Press, Axel Springer, and others in 2025
  • Announced a content partnership with The New York Times in September 2025
  • However, ignored negotiation proposals from Korea's three broadcasters

Diffusion Mechanism: Why File the Lawsuit Now

The Korean broadcasters' lawsuit is not an isolated incident. AI copyright litigation has exploded worldwide since 2025:

United States

  • September 2025: Anthropic settles writer class action lawsuit for up to $1.5 billion (Bartz v. Anthropic)
  • August 2025: Eight newspapers owned by Alden Global Capital sue OpenAI and Microsoft
  • 2024-2025: Major media outlets including The New York Times and Getty Images file lawsuits

Europe

  • 2025: German music copyright organization GEMA wins court victory against OpenAI
  • EU AI Act (effective 2024) mandates transparency for AI training data

Gap in Korean Law

While Korea enacted the Artificial Intelligence Basic Act in 2024, there are no clear provisions regarding copyright in training data. According to OhmyNews analysis:

  • No clear criteria for whether AI training constitutes "fair use"
  • Copyright law amendments under discussion but passage delayed
  • This lawsuit is effectively Korea's first judicial determination opportunity

The Irony of the 2025 Seoul Visit

OpenAI CEO Sam Altman visited Seoul in February 2025, emphasizing cooperation with Korea's AI ecosystem. But one year later, Korean broadcasters are confronting OpenAI in court.


Stakeholders: Who's Involved

Plaintiffs: Korea's Three Terrestrial Broadcasters

KBS (Korean Broadcasting System)

  • Public broadcaster with over 70 years of news archives
  • Revenue: License fees + advertising

MBC (Munhwa Broadcasting Corporation)

  • Public broadcaster (70% government stake), over 60 years of history
  • Revenue: Advertising

SBS (Seoul Broadcasting System)

  • Commercial private broadcaster, over 30 years of history
  • Revenue: Advertising

All three networks invest massive costs in news production. Annual investments of hundreds of billions of won for reporters, filming equipment, and editing staff.

Defendant: OpenAI

  • Founded in 2015, opened the generative AI era with ChatGPT in 2022
  • 2025 valuation: $157 billion (approximately ₩228 trillion)
  • 2024 revenue: $3.4 billion (approximately ₩4.9 trillion)
  • Major investor: Microsoft ($13 billion investment)

Key Point to Watch: Other Media Outlets' Choices

Other Korean media outlets likely suffered similar damages. Depending on this lawsuit's outcome:

  • Possibility of additional class action lawsuits
  • Or full-scale paid licensing negotiations with OpenAI

Sustainability: How Long Will This Last

Lawsuit Duration Forecast

Given the nature of international litigation, at least 1-2 years expected:

  1. First instance verdict: Expected first half of 2027
  2. Appeals: High likelihood of appeal by losing party
  3. Final judgment: 2028 or later

Short-term Impact (6 months-1 year)

  • Whether Korean courts will grant preliminary injunction (demanding suspension of ChatGPT Korean service)
  • Possibility of settlement offer from OpenAI
  • Preemptive licensing agreements by other AI companies (Google, Anthropic, etc.)

Long-term Impact (2-5 years)

  • Acceleration of copyright law amendments (new provisions for AI training data)
  • Establishment of collective licensing system (music copyright association model)
  • Development of Korean AI ethics standards

Secondary Issues: Derivative Arguments

OpenAI claims transformative use as grounds for fair use in US lawsuits. However:

  • Korean copyright law's fair use provisions are narrower than US law
  • Large-scale training for commercial purposes unlikely to qualify as fair use

2. Borderless AI vs Territorial Law

ChatGPT is accessible worldwide, but copyright applies by country:

  • Can a Korean court judgment be enforced against OpenAI in the US?
  • International judicial cooperation needed

3. Public Nature vs Commercial Nature of News

News has strong public good characteristics, but production requires massive costs:

  • If AI unauthorized training is allowed → Media outlet revenue model collapses
  • If media outlets collapse → Quality news production impossible
  • Ultimately AI also faces depletion of reliable training data

4. Data Sovereignty and Digital Colonialism

Viewing Big Tech's unauthorized data collection as "digital colonialism":

  • AI trained on Korean content provides services in Korea without compensation
  • Need to establish data sovereignty

Risks: What to Watch Out For

Plaintiff-side Risks

1. Possibility of Defeat

  • Foreign precedents: Some courts have ruled AI training as fair use
  • If defeated: Litigation costs + precedent established → loss of negotiating power

2. Technical Proof Challenges

  • Difficult to prove exactly what data ChatGPT used
  • OpenAI refuses to disclose training data

3. Public Opinion Division

  • Some public opinion: "Excessive copyright claims hindering AI development"
  • For public broadcasters: Claims that "content made with taxpayer money should be shared"

Defendant-side (OpenAI) Risks

1. Monetary Damages

  • Anthropic case: $1.5 billion (approximately ₩2 trillion) settlement
  • Three Korean broadcasters' lawsuit also expected to reach tens to hundreds of billions of won

2. Restricted Access to Korean Market

  • If preliminary injunction granted, ChatGPT Korean service temporarily suspended
  • Opportunity for Korean AI companies like Naver and Kakao

3. Global Chain Litigation

  • If Korean broadcasters win, Japanese, Taiwanese, Southeast Asian media may also file lawsuits
  • "AI copyright war" spreads globally

Checklist: Response by Stakeholder

Media Outlets (Content Creators)

Monitor whether company content is being used for AI training
Establish strategy for joining class action or individual negotiations
Strengthen content protection technology (watermarking, robots.txt)
Explore win-win licensing models with AI companies

AI Companies

Closely review Korean copyright law and precedents
Preemptive licensing negotiations with Korean media outlets
Ensure transparency of training data sources
Build compliance system for global regulations like EU AI Act

Government/Legislators

Urgent need for legal reform regarding AI training data copyright
Consider introducing collective licensing system (copyright trust management)
Strengthen international judicial cooperation (ensure judgment enforceability)
Find balance between media industry protection and AI innovation

General Users

Critically evaluate sources and reliability of AI-generated content
Consider paid subscriptions to sustain quality journalism
Express interest and opinions on AI ethics and copyright issues

  1. Korea Herald (2026-02-23): "Korea's top TV networks sue OpenAI for training ChatGPT with their news"
  2. Korea Broadcasting Association (2026-02-23): Official announcement of three terrestrial broadcasters' lawsuit against OpenAI
  3. KBS World (2026-02-23): "S. Korea's 3 Terrestrial Broadcasters Sue OpenAI for Using News Without Permission"
  4. Korea JoongAng Daily (2026-02-23): "Korean broadcasters sue OpenAI alleging unauthorized use of content"
  5. OhmyNews (2026-02-23): "AI Copyright War Enters Korean Courts"
  6. Maeil Business (2026-02-23): "News Isn't Free… KBS·MBC·SBS Sue OpenAI"
  7. Anthropic Settlement Case (2025-09): Bartz v. Anthropic, up to $1.5 billion settlement
  8. German GEMA Precedent (2025): Copyright infringement victory against OpenAI

Image Sources

Images not secured: Court litigation materials and official OpenAI statements are access-restricted and copyright-protected, making direct image acquisition impossible. Alternative description: Press conference scene from Sam Altman's Seoul visit in 2025 is symbolic, but photos from Newsis and other media outlets cannot be embedded in the text due to secondary copyright issues.

Related Posts