ChatGPT, Gemini, and similar chatbots run on models trained on hundreds of millions of books, academic papers, and online articles, most of which authors did not consent to have used. That arrangement feels illegal, but the courts say otherwise.
The Anthropic fine was not a ban
Last year, Judge William Alsup ordered Anthropic to pay $1.5 billion in damages to a group of writers. The penalty looked like a victory for authors, yet the judge ruled the training itself was lawful. The fine stemmed from Anthropic’s use of pirated books pulled from illegal shadow libraries.
In his ruling, Alsup wrote that Anthropic’s large language models did not train to replicate or supplant the works. Instead, he compared the process to a writer studying literature to turn a hard corner and create something different.
Cathy Gellis on the ruling
Cathy Gellis, an attorney specialising in intellectual property, copyright and technology, told TechCrunch that the decision helps AI companies. She noted that a $1.5 billion fine is small for a firm projecting about $200 billion in annual revenue by 2028.
“I think it is generally good news for AI training that he looked at what was going on and really sort of thought it analogous to reading a copyrighted work as opposed to copying a copyrighted work,” Gellis said. “Copyright law hinges on copying, but it doesn’t hinge on using the work or experiencing the work, consuming the work, reading the work.”
Law that has not changed since 1976
Copyright law has not been updated since 1976. Judges must interpret guidelines from half a century ago when addressing questions that could shape the AI industry.
“Everybody is very worried right now because the law is all over the place, and it’s because of this question,” Jason Henderson, Senior Attorney and Founder of the IP & Media Practice at JWL International, told TechCrunch. “They know that the AI model has been trained on so much stuff, and the law has not really caught up to that question.”
What courts look for in fair use
These cases often hinge on fair use law, which allows the use of copyrighted material without permission for criticism, parody, education, and other purposes. Judges weigh the purpose and nature of the work, the amount used, and the impact on the market.
“Copyright is always about protecting and growing the market,” Henderson noted. “The courts are kind of all over the place in their reasoning [in AI cases]. What’s tending to win is if what you’re doing is you’re training on somebody’s property because your purpose is to directly compete, then the courts will frown on it… If what you’re doing is not going to compete, then the courts are tending to find ways that it will be okay.”
Thomson Reuters versus Ross Intelligence
Henderson referred to a case where Thomson Reuters sued Ross Intelligence for copying content to build a competing AI legal platform. Judge Stephanos Bibas wrote last year that Ross’s use was not transformative because it lacked a further purpose or different character compared to Thomson Reuters’s work.
The judge decided it was not fair use to train on Reuters’ content to create a new platform that would directly compete with it. While authors could argue that chatbots compete with them by using their works to generate synthetic books, that argument has not yet won in court.
Distinguishing training from output
Gellis finds it helpful to separate two questions: how copyright applies to training AI models, and how it applies to AI-generated content.
In Thaler v. Perlmutter, a court ruled that a work created 100% by AI is not copyrightable. This raises issues about proving whether a work was generated by AI and determining what percentage was created or assisted by the technology.
“If you write your novel in [Microsoft] Word and run spell check, we kind of feel comfortable with the idea of saying that Word does not own your novel,” Gellis said. “[AI] is forcing us to look at a whole bunch of decisions that we kind of ignored for a while.”
Why the situation remains uncertain
Most AI companies are still involved in pending litigation, meaning a definitive solution is not on the horizon.
“What you are seeing is that the initial opening volleys are being influential, and that influence itself could be undone if other courts decide different things, and it’ll take later states of litigation to figure out which one will prevail,” Gellis said. “But in the meantime, all these decisions are shaping everything that’s happening. It would be kind of foolish for the AI companies to ignore them.”
