Protecting Your Music in the AI Era with Source Packets

Jason Colapierto of Suede Labs AI

When I recorded my first song with the intent to publish it, I thought song rights began and ended with the U.S. Copyright Office. My next song had a co-writer, which brought new rights questions. Jason Colapietro, founder of Suede Labs AI, recommends drawing up agreements as if everything has already gone wrong. Suede Labs AI builds provenance tools for creators; Jason is also a guitarist, singer, and technologist.

Jason’s source packet is a per-song record of stems, versions, notes, texts, splits, and permissions that shows how a work came together. Songs get made in text threads, sessions, and DAW folders, and the record of who did what usually lives nowhere. Jason admits he’s not naturally organized, which may be part of why his system feels doable. We talk about what goes in a source packet, when to start one, how to handle splits made informally, and where the evidence runs out.

Source packet principles have always mattered, and AI is a new reason they matter even more. We’re in an age of AI-generated music where label lawsuits are turning into partnerships and regulation is unclear at best. Jason explains how AI and licensing changes make documentation more important, and why sync is where organized records help most.

Disclaimer: A source packet isn’t legal proof of ownership, and nothing in this conversation is legal advice.

Topics covered include what goes in a source packet, organizing past work, naming files, and adding metadata so stems make sense later, where to store a packet, and who gets a copy, what timestamps can and can’t do, a takedown story that cost one artist part of his own track, why sync licensors care about organized records, when a packet isn’t enough, and much more.

Never miss an episode