BTS and BLACKPINK Songs Flagged in K-pop AI Training Data Controversy

BTS and BLACKPINK have become part of a widening K-pop AI Training controversy after reported searches of The Atlantic’s public AI training-data tool found thousands of K-pop songs in datasets circulating among AI developers. Koreaboo reported on June 24, 2026, that searches across about 11 K-pop artists showed more than 2,000 songs appearing in the datasets, including more than 200 BTS tracks in one dataset and BLACKPINK among the major acts named in the broader concern. ¹

The findings add K-pop to a dispute that has already reached across pop, jazz, classical music, and independent catalogs. The Atlantic reported on June 14, 2026, that it had identified four large music datasets being shared among AI developers, including collections with about 12 million tracks, 9 million tracks, and two datasets with more than 100,000 tracks each. ²

BTS, BLACKPINK and the K-pop Dataset Findings

BTS 블랙핑크 AI 학습 Seoul music dataset controversy

The K-pop angle centers on search results surfaced through The Atlantic’s AI Watchdog database, a public tool that allows users to search music identified in the datasets. The Verge reported on June 20, 2026, that the database lets users look up music found in the identified AI training datasets, with three of the datasets distributed as links to songs on YouTube or Spotify. ³

Koreaboo’s report focused on the visibility of BTS tracks across several datasets and said one dataset alone showed more than 200 BTS songs. The same report said searches across about 11 K-pop artists found more than 2,000 songs, with BLACKPINK and other major acts included in the wider set of results. ¹

The reported findings do not, by themselves, establish how any specific AI product was trained or whether every listed song was actually downloaded and used by a commercial model. The available source material says the datasets are circulating among AI developers and that some contain links rather than audio files. That distinction matters because a dataset listing can point to music hosted elsewhere, while automated downloading is a separate step. The Verge noted that automated downloading can bypass platform systems intended to support creators. ³

Still, the appearance of BTS, BLACKPINK, and other K-pop songs in these searchable datasets is significant because K-pop catalogs are commercially valuable, globally distributed, and often controlled by major entertainment companies, labels, publishers, and rights holders. The available reports do not state whether BTS, BLACKPINK, their agencies, or their labels have issued specific responses to these dataset findings.

Wider AI Music Data Dispute

The K-pop findings sit within a larger debate about how music datasets are assembled and shared. Music Business Worldwide summarized The Atlantic’s reporting by saying that four datasets together contain more than 21 million recordings and have been downloaded several thousand times. It also said two datasets have public origins and were not created for commercial music-generator training, while the biggest dataset, LAION-DISCO-12M, contains more than 12 million tracks. ⁴

The Atlantic’s report said the datasets include major pop, jazz, classical, and lesser-known artists, showing that the issue is not limited to one genre or one market. It also said Google and Stability AI have described use of one Free Music Archive dataset in research papers. ²

The public search tool has already prompted reactions from musicians outside K-pop. Pitchfork reported on June 21, 2026, that multiple musicians responded after finding their songs in AI training datasets. SZA said 238 of her songs appeared in the data, while producer Kenneth Blume criticized AI music company Suno, saying, “I can’t imagine going into work daily knowing you are stealing from countless struggling musicians.” ⁵

Suno also responded to The Atlantic through spokesperson Rachel Racusen, who cited safeguards intended to protect against “unauthorized distribution, impersonation and manipulations.” ² The available source material does not include a direct response from Suno specifically addressing the K-pop search results.

Legal Context Around AI Training

The dispute is unfolding alongside litigation and platform-policy arguments about whether music uploaded to major services can be used for AI training. Pitchfork noted ongoing litigation involving Suno, Udio, and major labels. ⁵

A separate report from Digital Music News said Google filed a June 2026 motion to dismiss an artist-led lawsuit alleging that music uploaded to YouTube was used to train AI models such as Lyria without authorization. Google argued that YouTube’s terms granted it broad rights to use uploaded content in connection with its business, including AI training, while also disputing contributory and vicarious infringement claims. ⁶

That legal context helps explain why the dataset findings have become contentious. For artists and rights holders, the central concern is whether copyrighted recordings were collected, indexed, downloaded, or used to train AI systems without permission or compensation. For technology companies and researchers, the arguments often turn on dataset structure, platform terms, research use, safeguards, and whether specific model outputs infringe protected works.

The available reports do not resolve those legal questions for BTS, BLACKPINK, or any other K-pop act. They do show that K-pop songs appear in searchable data tied to the broader AI music training debate, and that the controversy is now extending beyond Western pop catalogs into one of the most globally visible music industries.

BTS BLACKPINK K-pop AI training data debate Seoul music scene

For now, the confirmed picture is narrower than the debate around it: searches of The Atlantic’s database reportedly surfaced more than 2,000 K-pop songs across about 11 artists, including BTS and BLACKPINK, while the underlying datasets form part of a wider dispute over millions of tracks being shared among AI developers. The next developments are likely to depend on rights-holder responses, litigation, and whether AI companies provide more detailed explanations of what music was actually used to train their systems.