The mainstream read on the new lawsuit from Sony Music, Warner Chappell, and other music publishers against Anthropic goes something like this: the old guard is panicking, throwing lawyers at a technology it doesn’t understand, trying to slam the brakes on AI. I don’t think that’s what’s happening at all. I think these companies understand exactly what AI is worth, and that’s precisely why they’re in court.
Let me explain what I mean, and why it matters even if you’ve never written a line of code.
What actually happened
Sony Music, Warner Chappell, and other publishers have sued Anthropic — the company behind the Claude AI assistant — alleging what they describe as a “brazen campaign” of intellectual property theft. The core claim is that copyrighted works were used illegally in training the AI. That’s the news, as reported by TechCrunch, The Economic Times, and others.
If you’re new to this space, the phrase “used in training” is doing a lot of quiet work in that sentence, so it’s worth unpacking.
How an AI “learns” from things it doesn’t own
An AI model isn’t a database. It doesn’t store a copy of every song lyric or article it has seen, filed away for later retrieval. It’s closer to a very elaborate pattern-recognition machine. During training, it processes enormous quantities of text, and it adjusts millions of internal settings based on what it sees. Words that tend to follow other words. Structures that tend to appear in certain kinds of writing. Rhythms and rhymes and turns of phrase.
The result is a system that can produce text that feels like it was written by a person, because it was shaped by text written by people. Lots of people. Possibly including people who never agreed to be part of it.
That’s the crux of the disagreement. AI companies have generally argued that training is a transformative use — that reading a lot of material to learn statistical patterns isn’t the same as copying it. Rights holders argue that if you fed copyrighted work into a commercial product without permission or payment, you used that work, whatever you want to call the process.
Why I think this is about money, not resistance
Here’s my contrarian take, and it comes down to how these companies behave when they’re actually threatened versus when they smell an opportunity.
Music publishers have been through this before. Napster. YouTube. Streaming. In each case the initial move was litigation, and in each case the endgame was a licensing deal. Sony and Warner did not shut down streaming. They took a cut of it, and streaming became the industry’s main revenue engine. The lawsuit is the negotiation.
Read the language they chose. “Brazen campaign” is not the vocabulary of a company trying to establish that a technology should not exist. It’s the vocabulary of a company establishing that a technology took something valuable without asking. Those are different arguments with different desired outcomes. The first ends in an injunction. The second ends in a contract.
What this means for you if you use AI agents
You might be wondering why any of this should concern you when you’re just asking an assistant your inbox. A few reasons.
- Cost. If licensing deals become the norm, the price of building and running these systems goes up. Some of that lands on users, eventually.
- What your agent can and can’t do. Legal pressure tends to produce guardrails. Tools may start declining requests that touch song lyrics, published books, or copyrighted styles — not because they can’t, but because someone’s counsel said no.
- Who gets to build. Large AI companies can afford to negotiate with major publishers. Smaller teams often can’t. Copyright fights have a way of consolidating an industry around whoever can pay.
The trust problem underneath all of it
This lawsuit landed in the same news cycle as a story about KPMG pulling a report on AI usage due to apparent hallucinations — an AI-generated document with fabricated content, from one of the largest professional services firms in the world. Meanwhile, OpenClaw creator Peter Steinberger joined OpenAI, another reminder of how fast talent and attention move in this field.
Put those together and you get a fair picture of where things stand. The technology is moving quickly enough that top builders are being absorbed by the biggest labs. It’s also unreliable enough that a major firm had to retract published work. And the question of what it was legally allowed to learn from is still unresolved.
That’s not a technology in crisis. That’s a technology being priced. My honest expectation is that this ends with money changing hands and Claude continuing to exist, somewhat more expensively, with a slightly shorter list of things it will discuss. Whether that’s a good outcome for the songwriters whose work is at the center of it is a separate question, and one the lawsuit won’t necessarily answer.
🕒 Published: