Sony's New Tech Could Make AI Music Pay the Original Artists

It takes 10 seconds for AI to create a Beatles-style song today. But who should get paid for it?

Sony Group has developed technology that can identify the original music embedded in AI-generated songs, potentially opening the door for songwriters to claim compensation when their work is used to train AI systems.

The announcement comes as the music industry grapples with a fundamental question: if AI learns from millions of existing songs to create "new" music, shouldn't the original creators get a cut?

The Tech: Musical DNA Detection

Sony's system works like a forensic tool for music, analyzing patterns in AI-generated tracks to trace back to their original sources. Think of it as DNA testing for melodies – identifying which existing songs contributed to an AI's creative output.

Currently, AI music generators like Suno AI and Udio train on vast databases of existing music without paying royalties to original artists. These platforms can churn out professional-sounding tracks in any style, from jazz to hip-hop, all while the original creators whose work made it possible see nothing.

Winners and Losers Emerge

The clear winners? Major record labels and established artists with extensive catalogs. Companies like Sony Music, Universal Music Group, and Warner Music own rights to millions of songs. If AI companies must pay for using this content, these labels could see substantial new revenue streams.

Sponsored

Advertise with Us

[email protected]

Sponsored

The potential losers? AI music startups operating on the assumption that training data is "free." Suddenly facing licensing costs could force business model overhauls or even shut down smaller players entirely.

Independent artists face a complex situation. While they might finally get compensated for AI training use, lesser-known tracks could be filtered out of training datasets entirely, reducing their already limited exposure.

The Compensation Conundrum

Here's where it gets tricky. If an AI trains on 10 million songs to create one new track, how do you calculate each original artist's contribution? Should someone whose 30-year-old B-side influenced 0.001% of an AI's output receive royalties?

The technology exists, but the economic framework doesn't. Unlike traditional sampling, where you can point to specific lifted sections, AI influence is diffuse and statistical. Sony's detection system might identify the "what," but determining the "how much to pay" remains an open question.

Industry Disruption Ahead

This development could reshape the entire AI music landscape. Platforms might need to:

• Negotiate licensing deals with major labels upfront
• Develop "ethical AI" models trained only on royalty-free content
• Pass costs onto users, potentially ending the era of free AI music generation

For consumers, this could mean paying subscription fees for AI music tools that were previously free, or seeing AI-generated music quality decline as training datasets shrink.