screw the money — anthropic’s $1.5b copyright settlement sucks for writers

Authors Reach $1.5 Billion Settlement with Anthropic

Approximately 500,000 writers are now qualified to receive a payment of at least $3,000. This is due to a significant $1.5 billion settlement reached in a class action lawsuit. The lawsuit was originally brought against Anthropic by a collective of authors.

A Landmark Settlement, But Not a Victory for Authors

This settlement represents the largest financial payout in the history of U.S. copyright litigation. However, it should be viewed as a benefit for technology companies rather than a win for the authors themselves.

Technology corporations are actively competing to gather extensive amounts of written content. This material is crucial for training their Large Language Models (LLMs). These LLMs are the engines behind innovative AI chat applications like ChatGPT and Claude.

The Drive for Data and the Issue of Copyright

The sophistication of these AIs increases with the volume of data they process. Having scraped a substantial portion of the internet, these companies are now facing a scarcity of new information sources.

Consequently, Anthropic, the developer of Claude, resorted to utilizing millions of books obtained from unofficial “shadow libraries.” These books were then used to train its AI model.

The case, known as Bartz v. Anthropic, is one of many legal challenges. Similar lawsuits have been filed against companies such as Meta, Google, OpenAI, and Midjourney. These cases question the legality of training AI systems on copyrighted materials.

Settlement Details and the Core Issue

The settlement isn't a result of the AI training itself, but rather a consequence of Anthropic’s unauthorized downloading of books. It represents a financial penalty for illegally acquiring content instead of purchasing the necessary licenses.

Fair Use Doctrine and the Court Ruling

In June, Judge William Alsup ruled in favor of Anthropic. He determined that training AI on copyrighted material is permissible under the “fair use” doctrine.

The judge reasoned that this application is “transformative” enough to be protected. The fair use doctrine, however, has remained largely unchanged since 1976.

“Anthropic’s LLMs, much like any aspiring writer who reads extensively, utilized the works not to directly copy or replace them, but to develop new and original creations,” Judge Alsup stated.

Piracy as the Catalyst for the Trial

It was the act of piracy, not the AI training process, that initially led Judge Alsup to schedule a trial. However, with Anthropic’s settlement, a full trial is no longer required.

Aparna Sridhar, Deputy General Counsel at Anthropic, released a statement. It indicated the settlement would resolve the plaintiffs’ remaining claims. Anthropic remains dedicated to developing secure AI systems.

These systems aim to assist individuals and organizations, promote scientific advancements, and address complex challenges.

Precedent and Future Cases

As numerous additional cases concerning the intersection of AI and copyright proceed through the courts, Bartz v. Anthropic will serve as a legal precedent.

Given the significant implications of these rulings, it is possible that future judges may reach different conclusions.