There's a quote going around. Naomi Klein, aimed at AI:
“You ingested the entire written output of human civilization without consent, without compensation, and without credit, to build a system whose primary commercial application is eliminating the jobs of the people whose work you consumed. You are not liberating human creativity — you are strip-mining it and selling it back at a markup while calling the theft ‘training data." - Naomi Klein
It's a great line. It's built to be screenshotted.
And it falls apart the second anyone who actually understands the internet, AI tech, the law, or how creativity works.
Three claims. Let's kill all three.
## 1. "Without consent."
When you post something publicly on the internet, you consent to it being viewed, indexed, crawled, cached, and learned from.
That's not my opinion — that's how the web has worked since the day it was built.
It's called **implied consent**, and it's a real legal doctrine. You put your work in the public square. Now you're shocked the public viewed it, downloaded it, shared it and interacted with it? You don't get to expect your content to be discoverable and then cry "theft" when something discovers and uses your content.
But notice what creators aren't actually angry about. Not the viewing. Not the indexing. Not the crawling or the caching. Every one of those happens a billion times a day and nobody writes a manifesto about it.
The whole complaint hangs on one phrase: Trained on. That's the word doing all the work, and it's the exactly where the argument falls apart.
Because human beings train on data the same way. Every creator that has ever existed has spent years — decades, in a lot of cases — consuming content. Images, music, writing, video. Books, magazines, movies, TV shows, concerts. Every phrase, every quote, every novel they've read, every comic book, every song they've ever heard, every riff they've ever caught on YouTube. Everything they've ever seen or heard shapes what they create. Art isn't created in a vacuum. Your brain takes it in, compresses it, stores it, and remixes it into something new — and you call that inspiration. A model does the same thing with mathematics instead of neurons. It's the same process. AI learns through pattern matching, just like the human brain — it's modeled on the brain, and it thinks, learns, and trains the same way we do. Pattern recognition is the heart of it. The only real difference is that AI is better, faster, and cheaper than we are. And that's the real rub.
And this isn't some new thing AI invented. Every platform that operates in public already runs on it. Search engines crawl, index, and learn from public data. Recommendation systems learn from behavior. News aggregators ingest and sort the entire web. That kind of use — machines learning from public data — has been tested in court and upheld for decades. AI isn't doing anything the internet hasn't done since it existed. And it isn't doing anything human beings haven't done since the first one of us copied the cave painting next door.
So consent isn't the question. Implied consent is real, and posting publicly grants it. The real question — the one Klein never asks, because there's no good answer — is this: why should AI be held to a standard we apply to no other system that learns from public data, and no human being who learns from that same data?
## 2. "Without compensation."
Watch the sleight of hand here, because it's the whole game. Posting publicly implies consent — yes. It does not imply a paycheck. Those are two different things, and she welded them into one sentence hoping you wouldn't notice.
When you put your work online, you give permission — to everyone, people and machines alike — to view it, scan it, cache it, index it. That permission is real and you granted it freely. But nothing in that act obligates anyone who learns from your work to pay you for it. There is no implied compensation. There never was. It wasn't part of the deal, because there was no deal — you posted it for free.
"Public" means accessible. It does not mean billable. You can put your work in front of the entire planet at no charge — which is exactly what posting publicly is — without the planet owing you a cent. Want to get paid? That's a separate transaction: a license, a sale, a contract you negotiate up front. You don't get to give it away and then mail an invoice because somebody learned something.
## 3. "Without credit."
Credit attaches to **republication.** That's it. If you reproduce someone's work — quote it, reprint it, copy it — you credit it, either with permission or under fair use. That's copyright law.
Learning from something is not republishing it. An AI model does not store your painting and spit it back out. It learns patterns — this is what a cat looks like, this is how a sentence flows — the same way a human brain compresses everything it's ever seen into instinct. Nothing is copied. Nothing is retrieved. Nothing is republished.
So here's the question for every artist demanding credit: do **you** credit every artist who ever influenced you? Every song you absorbed, every book that shaped your voice, every painting that taught your eye? Of course not. You can't. Nobody can. Because that's not how creativity works, and it never was.
## The part nobody wants to say out loud
Nothing is created in a vacuum. Nothing you've ever made is entirely original. You are the sum of your influences — every image, every sound, every word you've ever taken in. That's your training data. You remix it, recombine it, and call the result an idea. That's not an insult to creativity. That **is** creativity.
AI did nothing a human doesn't do. It looked at public work and learned from it. The only difference is scale — it did in bulk what we do slowly and call inspiration. That's the entire crime. Speed.
I'm not saying this from the cheap seats. I've been an artist and a writer for thirty years. AI flattened a business I built. And I'm not out here whining about consent I already gave or compensation I was never owed. I picked up the tool and went back to work.
Because here's what I've noticed: the loudest voices screaming "theft" usually aren't creators. They don't understand the technology, they don't understand the law, and they don't have skin in the game. They're scared, they're angry, and they've found a narrative that feels righteous. That's not advocacy. That's a mob.
## And the fight's already over
None of this is even new. Google built an empire by scanning the open web — and the world's books — to train its search engine on data it paid nobody for. Authors sued. It climbed all the way up. The courts ruled that scanning copyrighted work to build something new and transformative — a searchable index — was fair use, and the Supreme Court declined to touch it. Settled.
That is the same mechanism, bolt for bolt. Scan public data, learn from it, build a tool that sends people back out into the world. If indexing the sum of human writing to power search was legal — and it is, decisively — then learning patterns from that same public data to power a model stands on identical ground. You don't get to bless one and call the other a heist. It's the same machine.
The fight she wants to pick was already lost — a decade ago, by the exact people now making her argument.
Consent. Compensation. Credit. Three pillars, and every one of them collapses the moment you apply an actual fact. What's left isn't an argument.
It's rage bait. And it deserves to be called what it is.
Comments
Log in to comment