|
Apple has proposed paying around $50 million (approximately 651 billion Korean won) to some media outlets as compensation for using their content in generative AI training. Big tech companies, facing criticism for unauthorized use of content in AI training, have shown a trend towards paying fees for content use. However, in Korea, there is a lack of consensus on compensation for content use.
According to The New York Times on the 22nd, Apple has initiated negotiations with major media and publishing companies to use news content in generative AI training. The entities contacted by Apple include Conde Nast, publisher of magazines such as Vogue and The New Yorker, NBC News, and the magazine People. Industry sources indicate that Apple has proposed paying a minimum of $50 million as compensation for using content. This suggests Apple’s plans to extensively utilize texts, images, and other content owned by media outlets in their AI development in the future.
Not only Apple but also OpenAI signed a news usage fee agreement with Axel Springer, a German media group that publishes the political news outlet ‘Politico,’ on the 13th. The estimated scale of the contract is in the range of millions of euros annually. OpenAI has been investing in training AI with as much content as possible, having previously entered into contracts with entities such as the Associated Press (AP) and the American Journalism Project.
An industry insider stated, “The competitiveness of big tech companies, which are staking their future on AI, comes from training AI with high-quality learning data as much as possible.” They explained, “As criticism has intensified recently, accusing big tech companies of unauthorized use and training of content, development companies seem to be securing legitimacy by paying fees to media outlets and utilizing content before training AI.
Despite ongoing controversies over global big tech companies’ infringement of intellectual property rights related to generative AI since the emergence of OpenAI’s ChatGPT, there has yet to be a substantial agreement on content compensation for developers in South Korea. Naver, for instance, faced criticism for allegedly training its HyperCLOVA X without the consent of content creators, using news data, blog posts, and cafe articles. In response, the Korea Newspaper Association expressed its position on preventing copyright infringement in generative AI, stating that generative AI is learning news content, generating results without permission, and proper citation.
The Ministry of Culture, Sports, and Tourism has announced its intention to release ‘AI Copyright Guidelines’ by the end of the year. However, these guidelines are expected to focus on whether to recognize the copyright of content created by generative AI, such as text, images, and videos. Professor Choi Byung-ho of Korea University’s AI Research Institute pointed out that the interpretation could vary depending on the focus on public interest or intellectual property protection. He stated, “If generative AI is developed solely for the public good, developers should not seek commercial gains. However, since they will undoubtedly profit through services like API, they should provide appropriate compensation by sharing profits with data copyright holders.
By. Min Joo Kim
Most Commented