Musical AI raises $4.5M to track copyrighted works in AI training
Musical AI, a rights management platform that enables tracing and licensing of copyrighted works used in generative AI training, has raised $4.5 million in funding to expand its attribution technology. The round was led by Heavybit, with participation from BDC and Build Ventures, the company announced this week.
​

The Halifax-based startup, formerly known as Somms.ai, develops technology that identifies which training data influences specific outputs from generative AI models and calculates what percentage of a generated output came from each source. The company said it will use the capital to grow its team, refine its attribution system, and establish new partnerships in the music industry before expanding into other creative sectors.
​

Published
Jan 13, 2026
musicbusinessworldwide.com
Claiming Attribution Is Possible at Scale
"Some claim attribution, licensing and AI are incompatible, or that only the largest players in the business can deploy it due to the cost and complexity," said Sean Power, CEO and co-founder of Musical AI. "We have proved them wrong. We have made attribution simple and turnkey."
​

The company is among the first to receive certification from Fairly Trained, a nonprofit founded by former Stability AI executive Ed Newton-Rex that evaluates and certifies AI models based on their training data sources. Musical AI has secured partnerships with audio content providers including Pro Sound Effects, SourceAudio, and Symphonic Distribution.
​

Jesse Robbins, General Partner at Heavybit, described the technology as "essential infrastructure that will enable and accelerate every media-focused AI product." He added that the platform offers AI companies "a seamless way to properly license, train, and use content while ensuring creators are credited and paid properly."
​

musicbusinessworldwide.com favicon
techobserver.in favicon
completemusicupdate.com favicon
5 sources
Building Licensed AI Music Tools
In December 2024, Musical AI partnered with India-based Beatoven.ai to develop what both companies describe as the "first fully licensed, rightsholder-compensating, generative AI platform" trained on copyrighted music. The platform, built on a dataset of over three million licensed songs, loops, samples, and sounds, is expected to launch later this year.
​

SoundBreak AI, formerly known as SESHY and led by Kevin Griffin of rock band Better Than Ezra, has also integrated Musical AI's platform. "Our business is built upon using licensed data sets with attribution for our AI model. Working with Musical AI is a perfect fit," Griffin said.
​

betakit.com favicon
beatoven.ai favicon
musicbusinessworldwide.com favicon
4 sources
Industry Context
The funding arrives as the broader AI music sector shifts from litigation toward licensing agreements. Warner Music Group reached settlements with both Suno and Udio in late 2025, while Universal Music Group settled its case against Udio in October 2025.
​

Power said Musical AI aims to demonstrate that licensing frameworks can scale alongside AI development. "We can not only license IP but also pay all involved rightsholders accurately and consistently," he said. "This ensures the future of human creativity will be enriched, not undermined, by AI."
​