South Korea is taking a leading role in regulating artificial intelligence, with the Korea Minting and Security Printing Corporation (KOMSCO) now commercially deploying technology designed to authenticate AI-generated content. The move comes as concerns mount globally over the proliferation of deepfakes and the potential for misuse of generative AI, and follows the enactment of South Korea’s comprehensive AI laws in January .
KOMSCO’s technology, a “non-visible digital watermark,” embeds a security code into images created by AI without affecting their visual quality. This allows for instant verification using a dedicated tool, confirming whether an image was indeed generated by AI. The technology is designed to be more secure than visible watermarks, and to deter attempts to conceal the AI origin of content. Crucially, the watermark persists even when the content is printed, extending its application to packaging, labels, and other physical products.
The initial commercial application of this technology is within LOOK PICK AI, a service that generates AI model images from user-submitted clothing photos. This platform, experiencing rapid growth, is particularly vulnerable to issues surrounding the unauthorized use and replication of AI-generated images. By embedding KOMSCO’s watermark, LOOK PICK AI provides a “digital authenticity certificate” for its output, allowing users to prove legitimate creation and assert ownership against potential infringement.
This development is directly linked to the core principle of South Korea’s new AI Basic Act: ensuring users are aware when content is AI-generated. KOMSCO’s technology provides a technical solution to meet this legal requirement. The company frames its role as extending the trust established through its traditional work with currency – a symbol of national confidence – into the digital realm.
The implications extend beyond simply identifying AI-generated content. KOMSCO’s technology aims to protect digital assets by establishing a clear record of origin and ownership. This is particularly relevant as AI-generated content increasingly becomes a commercial product. The ability to definitively prove authenticity is expected to be crucial for businesses operating in this space.
KOMSCO plans to expand the availability of its watermarking technology to other industries with high AI utilization, including webtoons and entertainment. This broader rollout aims to foster a more trustworthy AI ecosystem. The company’s technology research director, Lee Jong-sun, stated that KOMSCO’s role is to “implant the value of trust built through currency into the AI world,” and to create a “digital safe zone” where consumers can confidently interact with AI-generated content.
The move by South Korea comes amid growing global scrutiny of AI-generated content. In January , Elon Musk’s Grok chatbot faced investigations in India, France, and Malaysia over the creation of nonconsensual sexualized images. India’s Ministry of Electronics and Information Technology issued a 72-hour ultimatum to X (formerly Twitter) to address the issue, threatening the platform’s legal protections. France’s prosecutor’s office also launched an investigation into the spread of deepfakes on X, potentially violating the European Union’s Digital Services Act. Malaysia’s Communications and Multimedia Commission is investigating similar misuse of AI to manipulate images of women and minors.
These incidents highlight the urgent need for robust safeguards against the malicious use of AI. While the focus has been on preventing the creation of harmful content, the issue of copyright and intellectual property rights surrounding AI-generated works is also gaining attention. A recent LinkedIn post by Nikhil Narendran argued against regulating AI content generation based on similarity to existing works, suggesting that laws should focus on the dissemination of such content and hold human actors accountable. This perspective emphasizes that the act of “thinking” – whether human or machine – should not be subject to legal restriction.
The debate over copyright extends to trademark protection in the metaverse and other digital environments, as highlighted by recent research. The Korean Intellectual Property Office (KIPO) updated its trademark examination guidelines in to recognize the provision of digital goods online, signaling a growing awareness of the need to protect intellectual property in the digital realm.
South Korea’s proactive approach, combining comprehensive AI legislation with a practical technological solution like KOMSCO’s watermarking system, positions the country as a potential leader in establishing a framework for responsible AI development and deployment. The success of this initiative will likely be closely watched by other nations grappling with the challenges and opportunities presented by generative AI.
