AI Copyright's Unlikely Civil War: The People Who Hate IP Law Are Defending It

There is a contradiction hardening at the center of AI and the law, and it has nothing to do with which side is right. The same voices that spent years arguing intellectual property law was a corporate capture mechanism — a way for rights-holders to extract rent from culture without contributing anything back — are now its most passionate defenders. And the companies that built their reputations on openness, on remix culture, on the idea that information wants to be free, are now claiming that extracting knowledge from their outputs constitutes theft. The irony isn't lost on the people watching this unfold. "Isn't it ironic as hell," one observer put it on Bluesky, "that 'distillation' is accused of stealing intellectual property, while the AI companies themselves have used intellectual property from all over the world, incl. copyrighted stuff 'for free' to train their models?"[¹] It's a clean formulation of a genuinely messy problem, and it's the kind of observation that travels because it doesn't require you to pick a side to see the absurdity.

The legal machinery is catching up, slowly and unevenly. AI-hallucinated court filings are appearing with regularity now — the kind of professional embarrassment that a year ago read as a cautionary tale and today reads as industry weather. Meanwhile, the conversation in legal circles has shifted toward Section 230, the thirty-year-old liability shield that was written for a world where platforms were conduits, not authors. Generative AI blurs that distinction in ways the law has no clean answer for, and Congress is circling the question with the particular anxiety of people who understand the stakes but not the technology. A draft bill to revamp the online landscape is drawing attention in news coverage not because anyone thinks it will pass cleanly, but because the conversation itself signals that the old framework is no longer sufficient.[²] The question isn't whether Section 230 survives contact with generative AI — it's whether anything coherent replaces it before the litigation does.

The copyright question has its own specific gravity right now, and it's pulling in unexpected directions. The communities most skeptical of intellectual property maximalism are finding themselves arguing for stronger protections when the extracting party is an AI company rather than a major label or a film studio. That reversal isn't hypocrisy — it's a signal that what people actually care about isn't the abstract architecture of IP law but the power dynamic it encodes. When a small illustrator's portfolio gets scraped to train a model that then undercuts her rates, the argument "copyright is corporate capture" stops feeling like a liberation and starts feeling like a cover story. France's competition regulator has already moved — Google was fined for copyright violations related to its Gemini AI tool[³] — and the fine's scale, at $271 million, suggests that European regulators are treating this as a revenue question, not just a principles question. The creative industries are watching that number and doing math.

What's still largely unresolved is liability — who gets held responsible when AI causes harm, and under what legal theory. The Grok photo scandal generated enough alarm that a Chicago-Kent law professor was called in to explain it to a general audience[⁴], which is itself a kind of indicator: the law is moving from specialist debate to public explanation, the phase that typically precedes legislative action. The harder version of the liability question — not "who pays when the AI files a bad document" but "who is responsible when AI systems make decisions that harm people at scale" — is moving through courts slowly and through regulatory conversations even more slowly. The liability question has a way of surfacing everywhere and getting resolved nowhere.

The geopolitical dimension of AI law is adding a layer that domestic frameworks weren't built to handle. The US government's alarm about China's AI "distillation" — the practice of training models on the outputs of American AI systems — has been framed as an intellectual property emergency, but the underlying anxiety is about competitive advantage, not authorship. That framing matters because it pulls the legal conversation away from questions about creator rights and toward questions about national interest, which have different answers and different beneficiaries. Creators don't win when IP law becomes a trade weapon. They win when courts treat their work as something worth protecting in itself. Those two goals are currently occupying the same legal vocabulary, and the tension between them is going to force a clarification that nobody in the conversation seems eager to make.