AI company Anthropic settles with authors who alleged piracy

by

The tech company Anthropic, specializing in artificial intelligence, has resolved a class action lawsuit over copyright violations. Writers had claimed that Anthropic used their works to train its AI systems without obtaining necessary permissions, as outlined in a recent court document from last Tuesday.

In August 2024, the company Anthropic, located in San Francisco, faced a lawsuit filed by authors Andrea Bartz, Charles Graeber, and Kirk Wallace. This legal action came about due to the fact that their copyrighted works were utilized during the training of Anthropic’s AI assistant, Claude, based in San Francisco.

Details regarding the settlement agreement are confidential. The company has yet to provide a response following our request for a comment.

In June of last year, Judge William Alsup in the U.S. District Court made a ruling allowing AI models to study copyrighted literature without needing the writers’ explicit approval. This decision represented a moderate victory for Anthropic.

In the U.S. Court of Appeals for the 9th Circuit, Alsup determined that the usage was both “remarkably transformative” and fell under the category of a legitimate fair use. However, it’s alleged that the company may have violated copyright laws by illegally obtaining a substantial amount of its source material before it reached a settlement with the original authors. The tech company and the writers involved have requested the court to postpone proceedings until they finalize their agreement.

The “fair use” principle, enabling the copying of copyrighted works without permission under certain conditions, is an essential element in the defense strategies of AI companies when facing copyright infringement accusations.

Initially, Judge Alsup scheduled the case for a trial in December to determine the amount of compensation for copyright infringement. If the trial had taken place, each instance of willful copyright infringement could have resulted in damages as high as $150,000, potentially costing the startup billions. However, in early August, the AI company made an attempt to appeal the decision but was unsuccessful.

According to a judge’s ruling, Anthropic, an organization started by ex-employees of OpenAI and funded by Amazon, is alleged to have illegally obtained over 7 million books from online libraries like Books3, Library Genesis, and Pirate Library Mirror – platforms known for hosting unauthorized copies of copyrighted literature – in order to educate their software.

I was astonished to learn that they amassed vast quantities of physical books and meticulously dismantled them, digitizing each page for easy access. Intriguingly, a judge ruled that this practice fell under the umbrella of fair use, as determined by Alsup.

He pointed out that simply changing a printed book into a digital format, which allows for space saving and improved searchability, was already a significant transformation in itself.

Later, Anthropic, the company, bought the books that they had previously illegally copied. Alsup stated that these purchases wouldn’t entirely clear Anthropic of wrongdoing, but they could potentially lessen the amount of damages owed.

In a similar context, this agreement comes about due to an increase in copyright lawsuits against AI companies nationwide. For instance, Walt Disney Company and Universal Pictures have recently taken legal action against Midjourney, an AI company they claim utilized their copyrighted content for training image-generating models.

Read More

2025-08-27 01:01