I built this because I needed it.
How a musician using AI to create music about addiction and recovery discovered the provenance gap — and decided to close it.
By Abhi Basu · Founder, RightsDocket · March 2026
What is the provenance gap nobody talks about?
Every day, millions of creative works are produced with AI assistance. Music, images, text, video. The tools are extraordinary — they compress months of production into hours and make creative expression accessible to people who couldn't afford a studio session, a session musician, or a mixing engineer.
But the systems that prove authorship, establish ownership, and ensure regulatory compliance haven't kept pace. Not even close.
When you export a track from Suno or Udio, you get a bare MP3. No metadata about what's human, what's AI. No chain of custody. No way to register it correctly with the U.S. Copyright Office. The file lands on your desktop with all the provenance of a screenshot.
The performance rights organizations — ASCAP, BMI, SOCAN — now accept AI-assisted works. On the honor system. No cryptographic verification. No structured disclosure. No machine-readable metadata that a downstream platform, label, or sync agency can parse.
Meanwhile, the EU AI Act Article 50 enforcement date is August 2, 2026. Every organization producing AI-generated content will need machine-readable provenance metadata. Penalties run up to €15 million. California's SB 942 and AB 853 add U.S. transparency requirements starting the same year.
This isn't a future problem. It's a now problem with a hard deadline.
Where did this get personal?
I've spent 20+ years in regulated-industry communications — pharma, MedTech, medical devices. I know what it looks like when documentation systems can't keep up with the work they're supposed to govern. I've seen what happens when the audit trail doesn't exist, and the consequences land on the people who were just trying to do the work.
But it hit differently when it was my own music.
I started using AI tools to create music about addiction and recovery in the South Asian community. It's a topic that carries enormous stigma in our culture — something people don't talk about publicly. Music felt like the right way to open that conversation.
The creative process was genuinely collaborative. I wrote lyrics. I shaped melodies. I arranged sections. AI generated stems, suggested harmonies, handled production I couldn't do alone. The result was something neither of us could have made independently.
And then I tried to figure out: who owns this? Can I register it? How do I prove what I contributed?
The answer was: there's no standard tooling for any of this. The Copyright Office requires you to disclose AI contributions and file a “limitation of claim” — but there's no tool that maps your human contributions to their taxonomy. The C2PA content credentials standard exists, but zero AI music generators embed provenance metadata. The timestamping standards exist, but nobody has connected them to the creative workflow.
I was sitting at the exact intersection of the problem I'd seen in regulated industries for two decades and the gap I was now experiencing as a creator.
What did I build to close the gap?
RightsDocket is the platform I needed and couldn't find.
It combines three capabilities nobody else puts together: USCO Limitation of Claim automation that maps human vs. AI contributions to the Copyright Office's taxonomy, C2PA manifest generation producing machine-readable provenance metadata to spec 2.3, and RFC 3161 cryptographic timestamps that create proof-of-existence records that hold up in court.
Think of it as DocuSign for AI-generated content provenance.
The first product is the USCO Claim Generator — because the documentation layer is valuable today. Copyright registration doesn't wait for cryptographic infrastructure to mature.
C2PA-signed audio files (MP3, WAV, M4A) with ISRC/ISWC identifiers bound into the cryptographic envelope are next. That's the Audio Provenance Bridge — the missing layer between bare AI music exports and the music industry's metadata systems.
Why do the stakes demand action now?
The provenance gap isn't theoretical. These numbers tell the story.
until EU AI Act Article 50 enforcement
images watermarked via SynthID
AI tracks uploaded to Deezer
of online users seek "Verified" signals
tools combining USCO + C2PA + RFC 3161
You made something with AI. Make it yours.
Start with a free USCO claim preview. No credit card required.