Music Labels Propose Chart Restrictions on AI-Generated Recordings
Universal Music Group, Sony Music, Warner Music Group, and independent labels have proposed barring AI-generated recordings from official charts unless they are "substantially human made," a term left undefined. The proposal also requires lawful AI use and disclosure, but lacks operational definitions, verification mechanisms, and any commitment from chart operators to adopt it. IFPI's endorsement gives it international weight, yet it remains a recommendation with unresolved enforcement questions.

Listen to this dispatch
Narrated by an AI-generated voice.
Universal Music Group, Sony Music, Warner Music Group, and several independent labels have submitted a proposal to international charting bodies that would bar AI-generated recordings from official charts unless they meet eligibility criteria. The core requirement is that eligible songs be "substantially human made," a phrase the proposal leaves undefined.
The proposal goes beyond an earlier labeling initiative backed by the RIAA, the International Federation of the Phonographic Industry (IFPI), and SAG-AFTRA, which sought standardized tags distinguishing AI-generated from AI-assisted recordings. It keeps that labeling requirement and adds a chart-eligibility gate: recordings developed with generative AI could appear on charts only if the services used were lawful and authorized, the model had rights to its training data, the work was "substantially human made," and the release did not raise "stream or chart manipulation concerns." It also requires compliance with copyright, related rights, and personality rights laws, and says AI use must be signaled to consumers on streaming platforms.
The disclosure requirement aligns with earlier industry labeling efforts, but the chart restriction is the substantive change. Chart placement influences promotion, playlists, and industry leverage. A rule that excludes AI-heavy tracks would change commercial incentives for labels and artists considering fully or largely AI-generated releases.
IFPI has endorsed the proposal. No charting organization — such as Billboard or the UK's Official Charts Company — has said it will adopt the rules, so the proposal remains a recommendation from the recording industry rather than a binding standard.
Two definitions are unresolved. "Substantially human made" has no operational definition, and neither does the threshold of AI involvement that would disqualify a track. A song with AI-written lyrics but human performance, or human composition with AI-assisted production, falls into an unspecified gray zone. "Stream or chart manipulation concerns" is broad enough to cover everything from bot farms to promotional strategies that labels typically use, and the proposal does not explain what would trip that criterion.
The labels were asked for clarification. Sony Music, UMG, and Mom+Pop Music did not respond.
The proposal also does not address how compliance would be verified. If chart operators adopt these criteria, determining whether a model's training data was properly licensed, whether terms of service were breached, and what share of a recording is "substantially human" would require audit mechanisms that do not currently exist. Charting bodies would have to build them or rely on self-reporting from the labels submitting the music — the same labels proposing the rules.
The proposal echoes existing policies. The Recording Academy already requires "significant" human authorship for Grammy eligibility, and streaming platforms have their own AI disclosure policies. The chart proposal carries more weight because it comes from the labels themselves, which control most major releases, and because IFPI's endorsement gives it international reach. But without definitions, verification, or a chart operator willing to enforce the criteria, it functions for now as an industry position statement: the three largest music companies want AI-generated work kept out of the most visible measure of commercial success, and they have left the hard questions for someone else to answer.
Read the original at theverge.com →