Hello, this is Attorney Oh Seung-hyun from Cheongchul Law Firm.

In 2023, the Southern District Court of New York dismissed part of the copyright infringement lawsuit filed by famous authors against OpenAI. The authors claimed that "AI has unlawfully learned from our works," but the court found it difficult to conclude copyright infringement solely based on the similarity in the style of the outcomes.

However, this ruling does not justify the 'unauthorized data collection' by AI companies. The court clarified that “the scope and method of data collected for training AI models could pose issues,” distinguishing future liabilities. This is seen as a signal for AI companies to review the legality of their data collection practices.

AI learns from large data sets to infer patterns and generate outcomes. If it unlawfully collects and utilizes datasets from competitors, issues of ‘trade secret violations’ or ‘copyright infringement’ may arise. Particularly, if the outcomes provided by AI mimic the original data of competitors or resemble similar structures, stronger legal liabilities could follow.

Even if one claims simple 'open-source crawling,' the situation changes if the collected targets include competitors' APIs, news, customer reviews, etc. The legal ownership takes precedence over the justification of technology.

In January 2025, KBS, MBC, and SBS filed civil and criminal lawsuits against Naver, claiming that it unlawfully trained news content on its AI models. The original news articles are not only subject to copyright protection but also are major assets of the broadcasters. Notably, the fact that Naver trained this data into its proprietary AI news summary model became problematic.

This lawsuit is expected to be the first in the country to address whether ▲ the collection of articles for AI training purposes constitutes copyright infringement ▲ and whether it qualifies as trade secret infringement.

If Naver loses the case, domestic startups and medium-sized enterprises will also apply new standards for collecting AI training data. In particular, strategies for securing data through crawling may require a complete reassessment.

Currently, there is no special law regarding AI, but most judgments are made based on the following three general laws.

1.  Copyright Law: Using another person's text, images, or videos without permission constitutes copyright infringement, leading to a maximum punishment of imprisonment for up to 5 years or a fine of up to 50 million won (Article 136, Paragraph 1, Item 1 of the Copyright Law).

2. Unfair Competition Prevention Law and Trade Secret Protection Law: If one mimics or unlawfully collects and uses a competitor's database, algorithm structure, or API usage, punitive damages up to five times the amount of loss may be claimed (Article 14-2, Paragraphs 6 and 7 of the Unfair Competition Prevention Law).

3. Personal Information Protection Law: If data contains personal information (e.g., nicknames, location information in reviews) and is used without consent, a fine of up to 50 million won may be imposed (Article 75 of the Personal Information Protection Law), as well as administrative sanctions, such as business suspension.

If unauthorized usage of a competitor's data is acknowledged, damages can also be calculated for ▲ lost business opportunities ▲ and consequential damages due to decline in brand trust, in addition to simple compensation for losses (Article 125, Paragraph 1 of the Copyright Law).

Particularly, if 'intentionality' is proven, ▲ estimated damages based on sales, ▲ and the defense costs of competitors may also be included as subjects of compensation. If punitive damages under the Trade Secret Law are applied, one may have to pay up to three to five times the actual damages (Article 14-2, Paragraphs 6 and 7 of the Unfair Competition Prevention Law).

In fact, in 2020, a startup exposed competitor's member review data on its service after crawling it, which led to a ruling of about 200 million won in damages for 'unfair competition.'

Additionally, in 2022, an employee of a music platform was sentenced to 3 years and 4 months in prison on charges of using competitors' music on a USB for AI recommendation algorithm testing. Regardless of AI, unlawfully using 'competitors' assets' carries strong legal liabilities.

Therefore, when publishing a program that collects data through AI, going through the following procedures can be a way to prevent future damages.

1. Before collection or during the development stage: Conduct review meetings with external legal advisory teams at least once a month or have prior consultation through the Korea Copyright Commission (1800-1234) or the Ministry of SMEs and Startups legal support center -> It is particularly necessary to specify the method of API crawling/data processing.

2. Before the release stage: Certification of technological protective measures (TPM) or documentation of AI training data history (Trace) management.

Cheongchul Law Firm consists only of lawyers from Korea's top five large law firms, including Kim & Chang, Lee&Ko, Bae,Kim & Lee, Shin&Kim, and Yulchon, as well as legal teams from large corporations. Rather than just one lawyer, a team of specialized attorneys related to the case is formed to respond. Cheongchul provides comprehensive solutions for the entire business, not just solving specific issues, focusing on achieving what the client desires. If you need help achieving your goals, please do not hesitate to contact Cheongchul.