The Liancourt Rocks, known to Koreans as Dokdo, and to Japanese as Takeshima, are a pair of lonely, wind-swept volcanic islets in the waters between the Korean Peninsula and Japan. For decades, the territorial dispute between Seoul and Tokyo over these rocks played out through official statements and government posturing. But for South Koreans, Dokdo has never been just territory. It is a potent symbol of national liberation and the final clearing of the shadow cast by the Japanese colonial occupation (1910–1945). And today, that symbolism is no longer confined to state narratives – it is spreading across algorithmic feeds on TikTok and Instagram.
If you scroll through the “DokdoKorea” accounts across social media, you won’t find grainy archival footage or government-sponsored documentaries. Instead, you are met with the polished aesthetic of K-Pop. One particular track – mimicking the Oscar-winning song “Golden” – has already amassed over a million views on YouTube alone. The vocals are flawless, the hook is infectious, and synced perfectly with the high-energy beat.
But there is a twist: none of it is real. The singer doesn’t exist, the melody was composed in seconds, and the lyrics – rich with specific dates from 6th-century chronicles – were synthesized by artificial intelligence. This is the perfect example of digital nationalism with the usage of AI.
Across platforms like YouTube, Instagram, and TikTok, the “DokdoKorea” initiative has generated nearly 20 million views. By parasitizing global pop culture trends and the massive emotional reach of K-Pop fandoms, the creators are doing what state-led public diplomacy has failed to do for years: making a territorial dispute “viral” for younger generations.
This isn’t just a new form of fan-art; it is a sophisticated, low-cost, and highly effective model of grassroots and nationalistic politics of memory. In an era where generative AI can turn a historical grievance into a chart-topping bop, the battle for sovereignty is no longer just being fought with naval charts and exchanges of diplomatic statements – it’s being fought for “likes,” “views,” and the perfect digital “flow.”
This grassroots surge stands in contrast to South Korea’s traditional infrastructure of memory. For years, the state has institutionalized the Dokdo narrative through specialized law, museums, institutes, and school curricula.
The success of the “DokdoKorea” initiative is not a result of a sudden change in historical facts or changes in Korean memory, but a radical pivot in technological delivery. For eight years, the YouTube channel remained a relatively obscure corner of the internet, likely struggling with the high production costs and slow turnaround of traditional video editing and songwriting. That changed just months ago. By adopting generative AI tools for both music and visuals, the creators bypassed the bottleneck of human production, launching a digital blitzkrieg across TikTok, Instagram, and YouTube.
In this new model, the creation of nationalistic content is no longer a labor-intensive process reserved for state-funded broadcasters. Instead, it is a low-cost, high-frequency operation. The numbers speak for themselves: with just 20 full-length AI songs and a handful of Shorts, the channel has garnered over 8 million views on YouTube and next to 10 million views on TikTok and Instagram. The logic is simple: if the algorithm demands constant novelty, AI provides the infinite supply.
The idea of AI-generated tracks lies in their lyrical dissonance. While the music sounds like a chart-topping B-side from a Seoul production house, the lyrics function as a dense, rhythmic legal brief. These are not vague songs about “loving the motherland”; they are precise, pedagogical tools designed to weaponize history.
A recurring motif across the discography is a chronological “litany of proof.” In tracks that mimic the staccato delivery of modern K-rap, the AI-generated lyrics weave a tapestry of specific dates: 512 CE (the incorporation of islets into the Silla Kingdom), 1454 (King Sejong’s chronicles indicating Korean sovereignty), or 1877 (the Dajokan order, where Japan’s own Council of State admitted the islands were not Japanese territory). By embedding these dates into high-tempo hooks, the creators are effectively rote-teaching complex history to a wide audience, reinforcing the Korean narratives about the disputed islands.
Central to this grassroots narrative is the elevation of the “commoner-hero.” The songs frequently bypass high-ranking generals to focus on figures like An Yong-bok, a 17th-century fisherman who traveled to Japan to protest territorial incursions, and Hong Soon-chil, who led a volunteer civilian guard to protect the islands in the chaotic aftermath of the Korean War.
By framing the dispute through the eyes of the mincho (the common people), the AI lyrics create a direct lineage between these historical “volunteers” and the modern digital user. The message is clear: when the state’s diplomacy is perceived as too cautious, it is the ordinary Korean who must stand guard. This is the essence of grassroots politics of memory: the democratization of memory, where the “truth” is protected not by treaties, but by a collective, algorithmic “flow” that refuses to be silenced.
The AI’s lyrical scope also extends beyond the rocks themselves. Several tracks tackle broader “history wars,” criticizing Japan’s controversial textbook revisions and the visits of Japanese officials to the Yasukuni Shrine, which enshrines war criminals. By linking Dokdo to these larger issues of historical justice, the AI songs ensure that the territorial dispute remains inseparable from the broader demand for a “correct” reckoning with the past.
The emergence of AI-driven digital nationalism carries profound implications for East Asian stability. As for now, it’s just one channel, but it can be expected more AI-generated videos would be created. This phenomenon represents a significant shift in Korean collective identity. By democratizing the production of high-quality propaganda, AI can effectively take the monopoly on national memory away from the state and hand it to the Internet.
It could also change the relationship between voters and politicians. For a South Korean decision-maker, appearing “soft” on Dokdo is a perennial career-killer; now, that pressure is amplified by a 24/7 AI-generated soundtracks that frame any silence as a strategic defeat.
Furthermore, the “DokdoKorea” strategy exploits the inherent biases of social media algorithms. Platforms like TikTok prioritize high-engagement, emotionally charged content. A dry historical debate doesn’t trend; but a K-pop song with an “us-versus-them” narrative does. For Japan, this presents a “Whac-A-Mole” dilemma: how and if do you counter a decentralized, anonymous AI that can generate a dozen new songs for every one formal protest issued by the Ministry of Foreign Affairs?
The case is more than a quirky collision of high-tech and high-tension nationalism in East Asia. It might be a harbinger of a new era, where the ability to automate the cultural and historical legitimacy of a territory becomes a primary tool of non-state actors. As AI tools for music, video, and lyric generation become more sophisticated and accessible, this model is likely to be exported to other contested territories and historical disputes. In this environment, a single viral “song” can do more to cement a territorial claim in the minds of the younger generation than school lessons.
