IMPF and IMPEL have published a joint framework arguing that a composition’s lyrics and melody are at least as important to generative AI as the master recording, and should receive at least equal economic value unless other relevant factors justify a different allocation. Championed by IMPEL CEO Sarah Williams and IMPF President Annette Barrett, the initiative challenges what the organizations describe as a longstanding imbalance favoring master recordings over compositions in digital music revenues. Their framework proposes negotiating principles, not binding royalty rates.
The seven principles are:
- Value compositions properly across AI licensing models.
- Balance publishing and master revenues, starting from at least equal value for compositions.
- Distinguish payments for past use, training, outputs and future exploitation.
- Make deductions, costs and revenue calculations transparent and fair.
- Limit licenses to clearly specified uses, preventing unintended expansion.
- Agree attribution and valuation mechanisms with rightsholders.
- Provide downstream royalties for compositions used in model training.
The organizations also report three technology pilots with different companies, without identifying them individually. Their stated focus areas include:
- Identifying and tracking compositions through AI training.
- Attributing musical works across generation and outputs.
- Supporting reporting for accurate remuneration.
Join the discussion