A 1996 treaty grants performer rights but says nothing about AI

The audio task

A performer trying to understand what control they retain over a recorded performance, or a rights holder trying to work out what a streaming platform's on-demand access requires, is asking a question the WIPO Performances and Phonograms Treaty was built to answer during the analogue-to-digital transition of the 1990s, decades before anyone asked the same question about a synthetic voice.

What the documents show

The World Intellectual Property Organization's own treaty page states that the WPPT was adopted in Geneva on 20 December 1996 and addresses two kinds of beneficiaries: performers such as actors, singers and musicians, and producers of phonograms. The treaty's authentic text sets out specific rights rather than a general principle. Article 5 grants a performer moral rights: independently of economic rights, and even after those rights are transferred, a performer may claim to be identified as the performer of a live or fixed aural performance and may object to distortion, mutilation or other modification prejudicial to their reputation. Article 10 grants performers the exclusive right to authorize making a fixed performance available to the public in a way that lets listeners access it from a place and at a time they individually choose, the provision that later underpinned on-demand streaming licensing. Neither article, nor any other provision in the text, mentions artificial intelligence, synthetic performance or voice cloning.

Rights status

What the treaty grants is specific and dated: moral rights tied to a performer's own live or fixed aural performance, and an economic right over making that performance available on demand. What it does not do, because the question did not yet exist in 1996, is say anything about a performance that imitates a performer's voice without reusing their fixed recording. Extending the WPPT's language to synthetic-voice consent is an argument some advocates make today, not something the treaty text itself states.

What to check before you use it

This is an editorial checklist. Before citing the WPPT as authority for a voice-consent position, check whether the argument relies on the treaty's actual text or on an analogy to it. Confirm whether the jurisdiction in question has ratified and implemented the WPPT domestically, since the treaty binds contracting parties, not the industry generally. And separate a moral-rights claim, which concerns attribution and reputation, from an economic-rights claim, which concerns authorization and payment.

• Does the cited provision address a fixed performance, an unfixed one, or neither?
• Has the relevant country actually implemented the WPPT's obligations in domestic law?
• Is a synthetic-voice consent argument built on treaty text or on analogy to it?

The WPPT remains a foundational reference for performer rights precisely because it is specific about what it covers. Treating it as though it already answers the AI-consent question overstates a text that, by its own words, never reaches that question.