{"id":25407,"date":"2026-11-01T08:36:10","date_gmt":"2026-11-01T08:36:10","guid":{"rendered":"https:\/\/interspacemusic.com\/blog\/an-ai-disclosure-answer-is-not-a-remix-license\/"},"modified":"2026-11-01T08:37:48","modified_gmt":"2026-11-01T08:37:48","slug":"an-ai-disclosure-answer-is-not-a-remix-license","status":"publish","type":"post","link":"https:\/\/interspacemusic.com\/blog\/an-ai-disclosure-answer-is-not-a-remix-license\/","title":{"rendered":"An AI Disclosure Answer Is Not a Remix License"},"content":{"rendered":"

YouTube Music gives a delivered track three affirmative AI disclosure answers: fully generative AI, partly generative AI, or not generative AI<\/a>. Its other rule is just as consequential for catalogue work: an empty field means unknown<\/a>. A label deciding how a recording was made must answer that question separately from whether a platform may later let someone remix it.<\/p>\n

Make the disclosure an answer<\/h2>\n

\u201cUnknown\u201d is a distinct state, not a quiet version of \u201cnot generative AI.\u201d If a label leaves the field empty because it has not checked a track, the delivery carries uncertainty. If the label has checked and concluded that the track is not generative AI, it has an affirmative answer available. Those are different catalogue records, even when the audio file is the same.<\/p>\n

\n
<\/div>\n<\/figure>\n

The distinction matters most when several people have handled a recording before delivery. The person sending it to a platform may not be the person who knows how it was produced. A delivery choice made without that knowledge can turn a missing production record into a platform-facing answer. YouTube\u2019s disclosure decision travels through DDEX and CSV delivery<\/a>, so the practical control belongs before delivery: obtain the production account, choose the value it supports, and retain the account behind that choice.<\/p>\n

For an independent label, the useful unit of work is the track. Record who supplied the production information, what they said about generative AI, which answer the label selected, and when the answer was sent. If the information has not arrived, mark the decision for follow-up instead of treating a blank as a completed review. This gives the next person handling the catalogue a way to distinguish a deliberate answer from an unfinished task.<\/p>\n

Keep permission separate from production<\/h2>\n

Spotify is testing an AI-powered remix feature<\/a>, and an AI cover and remix tool could reach Premium subscribers in early 2027<\/a>. The participating artists have not been identified. That makes a rights decision different from YouTube\u2019s production disclosure: one describes the source track; the other would determine whether a new use of it is permitted and on what terms.<\/p>\n

A track marked \u201cnot generative AI\u201d could still be proposed for a remix. A track marked \u201cpartly generative AI\u201d still has a separate rights history to examine. Neither disclosure answer identifies who can authorize a cover or remix, what use an agreement covers, or who receives money from it. Treating the disclosure field as a permission switch would skip the contract question that controls the use.<\/p>\n

The money question is therefore concrete. If a remix uses a right controlled by an artist or label, their share depends on whether the relevant agreement recognizes that use and provides an accounting route for it. A term that permits use without payment to the controlling rightsholder would move the value of that use away from them. Before agreeing to participate, a rightsholder should ask what is being authorized, how the resulting use will be measured, and where any payment will appear. The unidentified participant list<\/a> makes those questions preparation for a possible offer, not a claim that a particular catalogue has already joined.<\/p>\n

Review the catalogue in two passes<\/h2>\n

A single \u201cAI reviewed\u201d flag is too broad to guide either delivery or negotiation. Use two passes with different outcomes:<\/p>\n