How the only U.S. organization collecting digital performance royalties is adapting its massive ISRC database to tackle deepfakes and ensure payment for featured and session artists.
Table of Contents
Unlike **ASCAP** and **BMI**, which collect royalties for the underlying musical composition (the song), **SoundExchange (SX)** is the sole organization designated by the U.S. government to collect and distribute **digital performance royalties** for the **Sound Recording** itself. [1]
SoundExchange is uniquely positioned in the AI conversation because its mission is tied to the master recording—the very asset most exploited by AI deepfakes and used for training. Its adaptation focuses heavily on data technology and protecting the rights of all performers on the track, including those who are often invisible.
SoundExchange collects royalties from non-interactive digital services, meaning platforms where the listener cannot choose the specific song on demand (acting like digital radio). These platforms include:
SoundExchange uses its patented matching technology to process **billions of performances** every month, ensuring accurate payment to artists and rights owners globally. [2]
The royalties SoundExchange collects are divided into three portions for every sound recording played, ensuring session musicians and background vocalists are compensated—a crucial element when discussing AI-cloned voices:
| Recipient | Share of Royalty | AI Implication |
|---|---|---|
| **Sound Recording Copyright Owner** (Label or Independent Artist) | **50%** | Determines the right to **reserve** the work from AI training. |
| **Featured Artist** (Lead Vocalist, Main Instrumentalist) | **45%** | The party most often impersonated by an unauthorized **AI deepfake** vocal. |
| **Non-Featured Artists** (Session Musicians, Background Singers) | **5%** | This portion is paid to organizations like **SAG-AFTRA** and the **AFM** for distribution, ensuring that every performer on the recording is paid for their contribution, whether or not the main artist consented to AI use. [1] |
SoundExchange is proactively developing a novel technical solution to the training data dispute: a specialized **AI sound recordings registry**. [4]
AI's high volume of content uploads exacerbates the music industry’s long-standing problem of inaccurate **metadata** (data about the song, artist, and composer splits). SoundExchange relies on perfect data to match performance plays to payments, and errors lead to "unclaimed" or "black box" royalties. [5]
SoundExchange is a founding member of the **Human Artistry Campaign**, a global alliance advocating for foundational principles regarding AI. This campaign emphasizes that: [6]
SoundExchange's response to AI is defined by proactive data management and strong advocacy. By developing an AI registry and leveraging advanced technology, the organization is attempting to provide a clear, efficient mechanism for rights holders to control their recordings' use in AI training. Its success in ensuring accurate metadata for AI-assisted works will be crucial for the financial health of featured and non-featured recording artists worldwide.
