Korea's new virtual idols debut before the rights are settled

Korea gained two new virtual idol groups inside six weeks. SEORIA, a four-member girl group from Giro, which operates the Dropshot AI platform, debuted August 31, 2026. muuo:23, a five-member boy group from Creation Jibup, is scheduled for October. Both were reported on September 9, 2026.

The pipeline behind them is not indie. SM Entertainment has developed the naevis character, Netmarble F&C and Kakao Entertainment created MAVE:, and JYP Entertainment has announced AI artist plans. Every major label in the country is now running a virtual-artist project, which means the category has moved past experiment and into slate planning.

The engagement numbers, read carefully

The figures being quoted for muuo:23 — roughly 48,000 pre-debut Instagram followers for one member, more than 138,000 likes on a fashion-brand advertisement featuring the group — are platform-reported and pre-revenue. They demonstrate that a synthetic act can accumulate an audience before it has released anything, which is genuinely a change from 2023. They do not demonstrate that the audience converts. The only Korean virtual act with a commercial record worth the name is PLAVE, and PLAVE's record is specific: a Melon top-100 entry, a sold-out Gocheok Sky Dome show in November 2025, and Billboard 200 and Artists 100 placements in April 2026. That took three years and human performers behind motion capture.

The distinction matters for anyone modelling this category. PLAVE is a costume around live performers. The newer acts lean harder on generative production — SEORIA was made without live-action shooting. Those are different cost structures and, more importantly, different labour structures.

Who owns what

That is where the business question sits. A virtual idol is at least three separable assets: the character design and its trademark, the voice — whether performed, modelled, or both — and the ongoing performance labour of whoever animates or voices it. Korean practice has not converged on how those are contracted, and the acts now debuting will set precedent by default rather than by design. We laid out the current state of ownership across the major groups in our reference on who actually owns Korea's virtual idols; nothing announced this month resolves it.

The pressure point is the human in the loop. When a group is a costume, the performer has leverage and a replaceability problem; when it is a model, the label has the asset and the performer has a licensing question. The same tension is playing out with licensed celebrity avatars, where the rights are at least negotiated with a named person who has counsel. A pre-debut synthetic act has no such counterparty, which makes the contracts cheaper and the eventual dispute harder.

Expect the next twelve months to produce more debuts than resolutions. The technical barrier to launching a virtual group is now low enough that label slate economics, not capability, decide the volume — and the rights framework will be written after the fact, in whichever dispute gets there first.