Hello. I am attorney Oh Seung-hyun from the law firm Cheongchul.

In the future, I will introduce the members of the Korean Artificial Intelligence Association to the conversations that take place at legal academic conferences or international seminars related to artificial intelligence, policy trends such as smart cities, and technologies produced by OpenAI and global big tech companies.

Technology always advances, but ultimately our laws, institutions, and social norms complete it. The technologies created by the members should also be within the boundaries of the law. I will emphasize again later, but technologists must be friendly with the law.

The story I want to discuss today starts from a very old question: “Isn’t it wrong to secretly use someone else's work?” This simple common sense permeates our entire law.

So how does the law facing artificial intelligence apply this common knowledge? Let's see if it presents new standards or simply recycles the old ones.

• AI Learning and Copyright: A Warning from the US Court

One day in 2023, the Southern District Court of New York was reviewing a lawsuit filed by famous authors against OpenAI. They claimed:

“AI secretly read and learned from our books. We never gave permission.”

The ruling led to the dismissal of some claims. The court determined that “the mere fact that the results generated by AI are similar to a writer's style does not automatically constitute copyright infringement.”

(Silverman v. Meta, Inc., No. 3:23-cv-03417).

From this perspective, it seems the law views the existing works and AI differently. However, this ruling did not grant a free pass to AI companies. The court states in the same ruling:

“The way and scope of data collection that an AI model uses for training can itself be problematic.”

Ultimately, the core issue was not whether the result was similar to a work, but how and what data was used to produce that result.

Meanwhile, in 2025, a US court ruled that the information collected for AI learning does not qualify as 'fair use' recognized under copyright law.

'Fair use' is a principle that allows unauthorized use of copyrighted works in specific circumstances, and it is also protected under domestic copyright law (Article 35-5 of the Copyright Act). However, with the US denying fair use, the likelihood of fair use claims being restricted domestically has increased. In particular, the chances of fair use being recognized are even lower when training on competitors' data.

Ultimately, the US acknowledged that far from granting a free pass for AI model training, copyright infringement can easily occur depending on the method and scope of data collection.

AI feeds on data. It becomes smarter and produces more sophisticated outputs. However, if that data comes from a competitor's API, news, customer reviews, or algorithm structure, the story changes. Can such acts be justified simply under the name of 'web crawling'?

• The Controversy of ‘Unlawful Learning’... The Boundaries of Copyright Facing AI

  
  

 Similar issues are also becoming apparent domestically.

In January 2025, in South Korea, the country's three major broadcasting companies—KBS, MBC, SBS—filed civil and criminal lawsuits against Naver.

The reason was that “they unlawfully used our news content for AI learning.”

The original articles are not just simple information; they are valuable assets of the broadcasting companies and edited creative works. If AI learns this data to create news summary algorithms, it could constitute copyright infringement or a breach of trade secrets.

Depending on the results of the above ruling, the scope of learning for Korean AIs may be significantly restricted.

Similar problems existed even before the emergence of AI.

In 2020, a startup crawled competitor review data and exposed it on its platform. The court deemed this an act of unfair competition, ordered the data obtained through crawling to be discarded, and ruled that the company should pay 20 million won to the affected party.

Ultimately, our legal precedents have recognized responsibility for the fact that “one unlawfully used another's assets” from the past, and it is expected to follow the same trend as the US ruling. That is, even though AI is an unprecedented technology, it is unlikely to serve as a shield for copyright law by its mere existence.

• AI Eats Data and Grows, But On the Table of Law 

The law has not yet established special rules solely for AI. However, general laws are already sufficiently strong.

· Copyright law states that anyone who reproduces or transmits another's work without permission can face up to 5 years in prison or a fine of up to 50 million won (Article 136(1) of the Copyright Act).

· The Anti-Unfair Competition Act allows for punitive damages of up to five times the loss if intent is recognized in the case of a breach of trade secrets (Article 14-2(6) and (7) of the Anti-Unfair Competition Act).

· The Personal Information Protection Act stipulates that fines and suspension of business can occur when using information such as nicknames and location information without the consent of the subject (Article 75 of the Personal Information Protection Act).

And as seen earlier, the courts (at least as of now) maintain the stance that existing works' standards and legal provisions should apply equally to AI. In other words, the belief that “data on the internet can be freely used” is a dangerous illusion.

AI endlessly desires more data. However, that desire must be contained within the vessel of responsibility. This is not merely a matter of obligation but a norm enforced by the judicial system, and non-compliance can result in civil and criminal penalties.

Our judicial system places emphasis on “how data is collected and used” when judging whether a violation of the law has occurred.

Managing the quality of goods is crucial in terms of costs. Therefore, the quality control of AI begins with excluding illegality. Ignoring this can impose unimaginable legal costs on developers.

Ultimately, AI must be friendly with the law, and developers handling AI need to maintain continuous communication with legal professionals.