Technology and Science23 August, 20263 min read
Judge William Alsup Rules Anthropic’s AI Training Lawful, Orders $1.5 Billion Settlement
This article is an automated aggregation of coverage from news outlets. Our editorials offer in-depth analysis curated by the NewsCord team.
Judge Alsup ruled Anthropic’s AI training on authors’ works lawful. Anthropic ordered to pay a $1.5 billion settlement.
Alsup ruling, $1.5 billion
US courts have not developed a unified approach to whether companies may train artificial intelligence models on books and other copyrighted works, and one of the first rulings came from Judge William Alsup in a case involving Anthropic. Alsup found Anthropic’s training of models on authors’ works lawful but ordered the company to pay $1.5 billion as part of a settlement over claims concerning the acquisition of books from illegal shadow online libraries.
TechCrunch reports that the judge distinguished between the training of large language models itself and the way the company obtained the books, comparing a model’s processing of a vast body of texts to a person reading literature to create something new rather than reproduce or replace the originals. Cathy Gellis, an intellectual property lawyer, said the court viewed model training more as becoming acquainted with a work than as copying it, and she tied the analysis to fair use factors including purpose and nature, amount used, and effect on the market for the original work.
Competing platforms, fair use
While Alsup’s ruling treated the act of training as lawful, the same legal framework produced a different outcome in the Thomson Reuters v. Ross Intelligence case, where Judge Stephanos Bibas ruled that Ross’s use of Reuters content to create a competing AI-based legal platform was not transformative and did not qualify as fair use.
Jason Henderson, senior lawyer and founder of the IP & Media practice at JWL International, told TechCrunch that courts tend to frown when a company “directly compete[s]” with the rights holder, but find ways for fair use to apply when there is no direct competition.
In the TechCrunch account, Henderson described the broader problem as a mismatch between how much data AI models absorb and the fact that “the law has not really caught up to that question.”
Cathy Gellis said the Alsup decision was “generally good news for AI training” because it treated AI learning as analogous to reading a copyrighted work rather than copying it, even as she emphasized that the law is “all over the place” in AI cases.
What’s at stake next
The legal uncertainty extends beyond training itself to the source of training data and the competitive effects of AI products, with TechCrunch noting that the Copyright Act of 1976 has not been updated since 1976 and courts must apply old rules to technologies that did not exist at the time.
UA.NEWS | Western Alternative says the principle of fair use is key in certain circumstances, but also stresses that courts consider the purpose and nature of the use, the amount of material used, and the effect on the market for the original work when assessing cases.
In addition to the Anthropic settlement, the sources describe a separate line of disputes over copyright in AI-generated content, including Thaler v. Perlmutter, where the court ruled that a work generated entirely by artificial intelligence cannot be subject to copyright.
TechCrunch frames the stakes as ongoing litigation and shifting boundaries, noting that “Most lawsuits against AI companies in this area are still pending,” while also pointing to the unresolved question of whether chatbots can directly compete with authors by generating synthetic books.