ASCAP put humans first in its AI principles

The tool moment

On 13 June 2023, the ASCAP Board of Directors adopted a set of AI principles, announced in a press release alongside a new AI Lab challenge, an AI panel at that year's ASCAP Experience, and a July AI Symposium. ASCAP frames the action on its own ASCAP & AI page with a quote from its CEO at that symposium: “Humans first. That is our position.” For a songwriter or composer member deciding how to think about a generative tool in their workflow, these principles are ASCAP's stated frame for how it will engage with the AI sector on their behalf, not a licensing rule with legal force of its own.

What the documents show

The press release and the ASCAP & AI page both list the same six principles: Human Creators First, prioritizing rights and compensation for human creativity; Transparency, in identifying AI-generated works and disclosing training data; Consent, protecting a creator's right to decide if their work trains a model; Compensation and Free Market, opposing government-mandated licensing that “would essentially eliminate consent”; Credit, where consent is granted; and Global Consistency across jurisdictions. The page's FAQ states directly that a member may not, at this time, register fully AI-generated works with ASCAP, but may register partially AI-generated ones. ASCAP is explicit that this reflects its own advocacy position, and separately notes that the US Copyright Office's own guidance agrees with its Human Creators First principle, a claim about alignment with a government document, not a claim that ASCAP's principles carry government authority.

What stays with the musician

ASCAP's principles are advocacy and internal policy, not a substitute for a member's own copyright analysis. A composer still has to determine, work by work, whether a piece meets the human-authorship standard the Office itself sets, since ASCAP's registration FAQ points members back to copyright.gov for that underlying rule rather than resolving it independently. Consent to AI training, in particular, remains something a member has to grant or withhold themselves; the principle states a preference for consent, not a mechanism that automatically enforces it.

Judge it by listening

There is no audio test for a set of principles, but the practical check for a member is procedural: before registering a work that involved a generative tool, can the human-created portion be separated out clearly enough to satisfy ASCAP's own partially-AI-generated category, or does the AI-generated share dominate the work in a way the FAQ says makes it ineligible altogether.

• Is this work partially or fully AI-generated under ASCAP's own definitions, and is it eligible for registration accordingly?
• Has consent been given, in ASCAP's sense, for any of this member's catalogue to be used in AI training?
• Does a claim that ASCAP's principles are legally binding overstate what a member advocacy position actually is?

A performing rights organisation's principles describe what it will fight for, not what current law already guarantees. ASCAP's own materials keep that line visible, pointing members to the Copyright Office for the legal question and reserving the principles for the advocacy one.