Copyright law hasn’t been this contested in decades. A wave of lawsuits against the biggest AI companies in the world is forcing courts to answer a question nobody fully settled before: is it legal to train an AI model on copyrighted work without permission? The answers so far are messy, contradictory, and worth billions of dollars. Here’s exactly where things stand.
A Record-Setting Settlement Just Reset the Stakes
The case everyone in publishing is talking about: a group of authors sued Anthropic in a class action, alleging the company illegally copied their books to train its AI models. After extensive briefing, the court ruled that training AI on copyrighted books can qualify as fair use, but storing pirated copies of those books does not. That distinction proved enormously costly — the case settled for roughly $1.5 billion, working out to an estimated payout of about $3,000 per affected work. It’s now considered the largest copyright settlement of its kind.
Two Courts, Two Different Rulings on the Same Legal Question
If you’re hoping for a clean, consistent answer on whether AI training counts as fair use, you won’t find one yet. One federal judge in a case against Meta granted a partial dismissal on fair use grounds for training its Llama model, while separately ruling that AI training “in many circumstances” would not necessarily qualify as fair use. Meanwhile, claims involving alleged reproduction of pirated works during the torrenting “seeding” process remain active and unresolved.
Google Just Got Hit With a Major New Lawsuit
The legal pressure keeps expanding to new companies. A group of publishers and authors — including major names like Hachette, Cengage, and Elsevier, along with author Scott Turow — recently filed a class action against Google, accusing the company of training its Gemini AI models on their copyrighted works without permission. The lawsuit goes further, alleging Google intentionally altered or removed copyright information to conceal that its models were trained on unauthorized material.
AI-Generated Art Still Can’t Be Copyrighted
One question did get a clear answer: can something entirely created by AI receive copyright protection? The Supreme Court recently declined to hear an appeal from an inventor seeking copyright registration for AI-generated artwork, letting stand a lower court ruling that upheld the Copyright Office’s refusal to register the work. The takeaway is clear — for now, U.S. copyright law still requires human authorship.
The UK and Germany Just Reached Opposite Verdicts
This is the detail legal analysts keep pointing to: the exact same underlying legal question produced completely different results in two countries within days of each other. The UK High Court ruled in favor of an AI company in a major image-rights case, while a German court ruled against an AI company on a strikingly similar fair-use-style question. It’s a clear sign that AI copyright law is going to look very different depending on which country you’re standing in.
The Music Industry Split Into Winners and Holdouts
Record labels have taken very different approaches to the same legal fight. Two major labels settled their AI lawsuits and signed licensing deals with AI music generators, effectively choosing to get paid rather than keep fighting in court. One major label, however, is still litigating against an AI music platform, declining the settlements its competitors accepted — a bet that the courts, not licensing deals, will deliver a better outcome.
Hollywood Is Suing Over AI-Generated Characters
It’s not just books and music. Major entertainment studios filed a copyright infringement lawsuit against an AI image-generation platform, alleging it allowed users to generate images of their copyrighted characters without authorization. It’s one of the clearest signals yet that the entertainment industry sees AI-generated content — not just AI training — as its next major legal battleground.
The Bottom Line
Copyright law is in the middle of its biggest identity crisis in a generation, with courts across the world reaching different conclusions on nearly identical questions. Publishers, authors, musicians, and studios are all fighting the same underlying battle from different angles — and with appellate courts still working through the first real test cases, the legal picture is far from settled. Whatever gets decided next will shape how creative work, and the AI systems trained on it, coexist for years to come.
Leave a Comment