I went through exactly this process.

Then I discovered (A)GPL and realized that the system makes sense to protect user rights.

And as I started making my own money, I started paying instead of pirating, though I sometimes wonder how much of my money goes to the actual artists and creators and how much goes to zero-sum occupations like marketing and management.

---

It comes down to understanding power differentials - we need laws so large numbers of individuals each with little power can defend themselves against a small number of individuals with large amounts of power.

(Well, we can defend ourselves anyway but it would be illegal and many would see it as an overreaction - as long as they steal only a little from each of us, we're each supposed to only be a little angry.)

---

> Maybe another 10 years and we'll be in the spot when these things are considered illegal again?

That's my hope too. But it requires many people to understand they're being stolen from and my fear is way too few produce "content"[0] and that the majority will feel like they benefit from being able to imitate us with little effort. There's also this angle that US needs to beat China (even though two nuclear superpowers both lose in an open conflict) and because China has been stealing everything for decades, we (the west) need to start stealing to keep up too.

[0]: https://eev.ee/blog/2025/07/03/the-rise-of-whatever/#:~:text...

just pirate again. It's the only way to ensure a game or movie can't be recalled by publishers the next time they want everyone to buy the sequel.

And that's just one part of it.

Part 2 is all the copyleft code powering the world. Now it can be effortlessly laundered. The freedom to inspect and modify? Gone.

Part 3 is what happens if actual AI is created. Rich people (who usually perform zero- or negative- sum work, if any) need the masses (who perform positive-sum work) for a technological civilization to actually function. So we have a log of bargaining power.

Then an ultra rich narcissistic billionaire comes along and wants to replace everyone with robots. We're still far off from that even if actual AI is achieved but the result is not that everyone can live a happy post-scarcity life with equality, blackjack and hookers. The result is that we all become beggars dependent on what those benevolent owners of AI and robots hand out to us because we will no longer have anything valuable to provide (besides our bodies I guess).

> Laws have to be enforceable.

This is a good point. In this case, it does seem pretty easy to enforce, though - just require anyone hosting an LLM for others to use to have full provenance of all of the data that they trained that LLM on. Wouldn't that solve the problem fairly easily? It's not like LLM training can be done in your garage (at which point this requirement would kill off hundreds/thousands of small LLM-training businesses that would hypothetically otherwise exist).

In most of those cases, it was because too many people broke the laws, regardless of what companies did. It was too distributed.

But to train a model, you need a huge amount of compute, centralized and owned by a large corporation. Cut the problem at the root.

Quantity matters.

Imagine it took 10^12 hours to produce the training data, 10^6 hours to produce the training algorithm and 10^0 hours to write a bunch of prompts to get the model to generate a useful output.

How should the reward be distributed among the people who performed the work?

>But why are you not talking about what the LLM user brings? They bring a unique task or problem to solve. They guide the model and channel it towards the goal. In the end they take the risk of using anything from the LLM.

I must remember next i'm shopping to demand the staff thank me when i ask them them where the eggs are.