Sony Music warns 700+ tech companies to stop scraping its songs for AI

If you're sitting on one of the richest catalogs in recorded music history, and thousands of AI crawlers are scraping it clean, at some point you have to slam the door.

Sony Music Group — the powerhouse behind Columbia Records, RCA, and Epic — just sent formal warning letters to more than 700 AI companies and streaming platforms. The message is blunt: stop using its music to train your models without a license.

Audio, lyrics, and even metadata are off-limits

Sony isn't leaving any legal gray area here.

The letter explicitly bars companies from web scraping, text mining, or data harvesting across a massive spread of assets. That includes master recordings, musical compositions, lyrics, album cover art, and even underlying metadata.

According to the notice, unauthorized training strips artists and labels of fair compensation and infringes on their intellectual property. Sony is demanding that these 700-plus recipients confirm they haven't touched its content, or provide full documentation detailing what was scraped and how it was used.

In a public statement, Sony acknowledged that tech disruptions are nothing new for the music business. But innovation, the label argued, cannot happen at the expense of artists' rights.

The legal heat is turning up

This isn't an isolated flare-up. The entire music industry has been pushing back against AI scraping for months.

Last year, the viral AI track "heart on my sleeve," which cloned the voices of Drake and The Weeknd, triggered swift takedowns by Universal Music Group. Drake found himself on the receiving end in April, pulling his track "Taylor Made Freestyle" after Tupac Shakur’s estate threatened legal action over an AI vocal clone.

Just last month, over 200 artists signed an open letter warning against AI that replaces human creativity, while Warner Music Group CEO Robert Kyncl testified before the U.S. Senate to push for federal deepfake protections.

Timing also matters. The EU passed its landmark AI Act in March, mandating that general-purpose AI developers disclose detailed summaries of their training data. Sony now has a very real legal framework to lean on.

Why it matters

For generative AI startups building music and audio tools, the days of "scrape first, apologize later" are officially wrapping up. If you want high-quality training data, you will have to license it, pay for it, or build without major-label sound.

We think Sony's move is calculated and smart. Rather than trying to fight AI development outright, Sony is defining the legal perimeter early — all the way down to metadata and cover art — to force AI builders directly to the negotiating table.

The label industry seems to be running the exact same playbook it used against early streaming platforms: threaten massive infringement liability until tech companies sign lucrative commercial agreements. The downside is that independent developers will almost certainly be priced out, leaving AI music creation in the hands of a few tech giants rich enough to pay the licensing toll.