Music Artists Should Be Paid.

A note from the studio on AI, ownership, and following the rules we already have.

Let's begin where it might surprise you to find us: music artists should be paid for their work. Fully, fairly, and as a matter of course.

We say "surprise" because we are an AI studio. BDL Music Studios makes music with these tools, openly and on purpose. The easy assumption is that an AI studio must sit on the other side of this argument — that we'd see royalties as friction and rights holders as obstacles.

We don't. We think the people who write and record music deserve to be compensated when their work is used, and we don't believe the arrival of a new technology changes that for a single second. What we want to push back on isn't the principle. It's the idea that AI requires us to invent a brand-new principle to enforce it.

The Rules We Already Have

Here is the thing the headlines tend to skip: music already has a mature, well-tested system for unauthorized use. It's called copyright law, and the industry has spent decades exercising it.

So our question is a simple one: why should using a more sophisticated tool change which rulebook applies?

"The wrong was never the tool. The wrong is taking someone's work without permission. A person who does that with a sampler and a person who does it with a model have committed the same offence, and the same law should meet them both."

Treating AI as a special case invites a double standard — and double standards rarely serve creators well in the long run.

Facing the Hardest Counter-Argument

Now the honest part, because we would rather meet the strongest version of the counter-argument than tiptoe around it. There is one place this genuinely gets difficult, and it's worth naming plainly: training data.

The hard question isn't whether a finished song copies another song — existing law handles that cleanly. It's whether the *models themselves* were built by learning from vast amounts of copyrighted music without permission. That is a real and unresolved debate, and the courts are working through it as we write.

But notice who can actually answer it. We can't. No end-user can.

When we sit down to make a song, we are paid users of legitimate, commercially-licensed tools — the same way a designer licenses stock images from Adobe, or a filmmaker licenses footage from a reputable supplier. How a given model was trained is knowledge held by the companies that built it, behind doors the rest of us will never see behind.

We made our choice in good faith: we use platforms that trade openly, charge for commercial rights, and have been tested in court. If there is a reckoning still to be had over training data, it belongs in the room where the knowledge and the power actually sit — between the rights holders and the AI companies. It does not belong on the desk of the musician who bought a licence and acted in good faith.

The Bottom Line on Legitimacy

We'll be fair to the other side here, because fairness is the entire point. "I just bought the tool" isn't a magic shield in every situation; knowingly buying something obviously stolen doesn't clear anyone.

But that is precisely why the legitimacy of the tools matters. These aren't back-alley operations. They are openly-trading, commercially-licensed platforms, several of them now in formal agreements with major labels. Buying a licence from a company operating in the open is not the same as receiving stolen goods, and it shouldn't be treated as though it were.

Where We Land: Our Four Standards

• ✔ Artists should be paid. Period.
• ✔ Unlawful use should be pursued — under the law we already have, not a hastily-built parallel one.
• ✔ Training-data accountability belongs with the companies that hold the answers, not the users who don't.
• ✔ The work itself should be judged on its merit, not on the tool that helped make it.

We are an AI studio that believes all of that. We don't think those ideas are in tension. We think they're simply fair.