You’ve probably noticed your feeds drowning in AI-generated images, articles, and songs. You might take comfort in the fact that, legally, a machine can’t own the copyright to any of it. You’d be wrong to feel safe.
The European Union just clarified something profound: copyright law protects human authors. Without a human creator, a work falls immediately into the public domain. On the surface, this feels like a victory for creators. A billion AI-generated books and images are suddenly a legal commons, free for anyone to use.
But look closer. This isn’t a shield for human creativity. It’s a massive, legally sanctioned loophole that rewards tech giants and encourages a new kind of fraud.
The law says a machine can’t own its output, but it says nothing about the machine eating our output to get there.
Think about the math. The same AI systems that cannot hold copyright were trained on a vast, scraped corpus of copyrighted human expression. We have built a one-way cultural relationship: AI can learn from everything we make, but the law ensures it gives absolutely nothing back. The machine takes our labor to generate its slop, and then dumps that slop into the public domain for its creators to monetize.
It gets worse. Because pure AI outputs are legally ownerless, anyone who wants to claim ownership just has to play a trivial game. I saw a commenter put it perfectly: “I will add a pixel by hand with software on my computer.”
That’s it. That’s the entire legal strategy. It’s called “human washing.”
A hand-placed pixel is the new signature on a forged masterpiece.
Anyone can generate an AI image, make a trivial edit, and claim sufficient human creativity to slap a copyright on it. This will produce a flood of weakly-authored, legally protected garbage that clutters the copyright system. Meanwhile, the pure AI outputs—the most scalable production model ever invented—remain free for massive corporations to exploit without paying a dime to the humans who fed the model.
For creators, this isn’t just an insult; it’s an existential threat to your income and attribution. For businesses, it turns every AI-generated asset into a potential liability. Are you buying a copyrighted image, or an ownerless public domain asset someone slapped a pixel on? You won’t know until you get sued.
We are watching the legal system shrug at AI as a creator, while it bends over backward to protect AI as a consumer of culture. The operative question is no longer “can an AI be an author?” It is now “how much human input is enough to claim authorship?”
Right now, the answer is terrifyingly low.
The machines don’t need to steal our copyright. They just need us to add a pixel and hand it over.
FAQ
Q: Doesn't this just mean everything AI generates is free for us to use?
A: No, because 'human washing' allows people to claim copyright on AI outputs by making trivial edits. The only things guaranteed to be truly free are pure, unedited AI outputs—which mostly benefit the corporations scaling the machines, not you.
Q: How much do I need to edit an AI output to claim copyright?
A: There is no bright-line rule, which is the problem. The law requires 'sufficient human input,' but a single edited pixel or a minor prompt tweak is currently being argued as enough to claim authorship.
Q: If AI can't own copyright, shouldn't AI be allowed to train on copyrighted material for free?
A: It's a fair-weather argument. Tech companies want a one-way street: free access to human copyright for training, but zero liability for the ownerless outputs they sell. You can't claim the rights of a consumer while dodging the responsibilities of a creator.