TL;DR  —  ONE-LINE SUMMARY

Studios have stopped watching from the sidelines: Disney and Warner Bros. are now suing Midjourney over AI-generated output that reproduces recognizable characters — not because of what AI trained on, but because of what it spits out. The lawsuits, ByteDance's viral deepfake scandal, and Disney's exit from its OpenAI deal are collectively drawing the first real legal boundary around IP reproduction in generative AI — and that boundary will reshape every content-tech partnership deal on the planet.

The Long Hesitation Is Over

For years, major Hollywood studios refused to sue AI companies. The reasons were pragmatic: litigation is expensive, copyright law for AI was unsettled, and — above all — studios did not want to permanently close the door on potential AI partnerships before those relationships had even begun. Even when the Writers Guild of America sent formal letters urging studios to take legal action over the use of Hollywood screenplays in AI training, the majors held back.

That restraint ended in the second half of 2025. Disney and Warner Bros. filed separate lawsuits against Midjourney, the AI image-generation service — and the legal strategy they chose signals something bigger than a one-off dispute.

"The target is not what the AI learned — it's what the AI shows you."

디즈니-오픈AI 결별:할리우드는 왜 이제야 AI를 고소하기 시작했나
스튜디오, 마침내 AI를 고소하기 시작. 훈련 데이터가 아니라 ‘아웃풋의 캐릭터 재현’을 겨냥한 디즈니·워너브라더스의 미드저니 소송은, 생성AI가 기존 IP를 어디까지 베낄 수 있는지에 대한 첫 법적 경계를 그리는 시험
K-EnterTech HubJung Han
Korean Version

What sets these Hollywood lawsuits apart from the music-industry copyright battles is where the legal crosshairs are aimed. Music-rights cases have focused largely on training data — whether AI models were built on copyrighted songs without licenses. The studios have chosen a different battlefield.