跳转到主要内容

此界面翻译为草稿,未经母语使用者审校。英文为源版本。 English

Suno and Udio roll out models trained on licensed music — but artist consent and pay remain unclear

2026年9月1日 · 自动从公开来源采集,未经人工审核。

这篇内容目前只有 英语 原文版本。

What happened

In September 2026, Suno rolled out its first models trained on licensed music — from Warner Music Group, BMG, and Believe — while rival Udio signed licensing deals with Universal Music Group and Warner. The deals mark a pivot from the major-label copyright lawsuits of 2024 (which have not fully ended: Sony and Universal are still suing Suno, per the Los Angeles Times) to a license-first model. But the terms are undisclosed: the LA Times reports none of the companies have disclosed what artists will be paid, how many have opted in, or what levers an individual musician can pull to keep recordings or compositions out. Because label contracts assign key rights, a label may license a master for training while the musicians who made it — session players, background singers, co-writers — have no practical veto. The piece notes the open technical problem of attribution: per-output royalties would require proving which training data shaped which output.

Why it matters to creators

This is the emerging blueprint for AI music compensation — labels licensing catalogs for training — and it reveals how little of it currently reaches individual artists. Musicians, especially contributors without approval rights, should understand that "licensed" does not mean "consented by the artist": the consent gap between labels and creators is the central unresolved issue, alongside undisclosed payouts and no audit rights on when works enter training.

What is not yet known

Deal terms, payout structures, opt-in numbers, and artist-level controls are all undisclosed. Litigation outcomes (Sony/UMG v. Suno) and the Copyright Office's AI policy study could still reshape the framework.

来源