Mixed Reactions Among Authors in Landmark AI Copyright Settlement

An AI company, Anthropic, has agreed to pay $1.5 billion in a copyright infringement settlement involving hundreds of thousands of authors. The reaction among authors is mixed, as some feel the outcome is not a complete victory.

RICCARDO MILANI/Hans Lucas/AFP via Getty Images/AFP hide caption toggle caption RICCARDO MILANI/Hans Lucas/AFP via Getty Images

Charles Graeber, a bestselling author and journalist, expressed pride in being one of the main plaintiffs in this significant case. “Joining a group of authors and publishers to challenge a company like Anthropic was a meaningful effort,” he shared with NPR. However, he hesitates to label the lawsuit’s outcome as an all-out win.

More than 300,000 writers could soon benefit from this settlement. A federal judge in San Francisco approved the settlement in July, following a class action lawsuit filed two years ago. The authors accused Anthropic of using copyrighted books to train its AI models without consent or compensation.

Graeber expects to receive around $3,100 for each of his two books used in training Anthropic’s chatbot, Claude. The verdict serves as a reminder that piracy remains a crime, but Graeber has mixed feelings about the personal and financial cost of the legal battle.

Bryan Bedder/Getty Images hide caption toggle caption Bryan Bedder/Getty Images

The $1.5 billion settlement, while substantial, must be shared among thousands of authors and their publishers. More than $100 million of the total went to the plaintiffs’ lawyers. Some view the arrangement as more of a win for AI companies, as the court ruled that using copyrighted books for AI training was not illegal, provided payment was made.

Aparna Sridhar, deputy general counsel for Anthropic, stated, “Training AI on books is fair use under copyright law.” Over 91% of the authors and publishers involved have claimed their payment share. Similar rulings have favored other tech companies like Meta in related cases.

A use of the fair use doctrine presents a challenge for human authors. Andrea Bartz, another plaintiff, emphasized that using algorithms to create texts that compete with human-authored works complicates the market.

International Concerns and Regulation

U.S. laws may not guarantee that other countries will follow similar copyright protections. Dale Nelson, an intellectual property lawyer, highlighted the threat from overseas AI companies, particularly from countries like China, where government control complicates enforcement efforts.

Bartz advocates for international cooperation in AI regulation. Chinese AI firms often use distillation techniques, which involve training their models on outputs from other AI, rather than directly on copyrighted texts.

Bartz remains hopeful about ongoing copyright cases involving AI in the U.S. courts. “This lawsuit could be the first step towards a fairer environment for creatives,” she said.

Licensing as a Solution?

Licensing offers a potential path forward for authors to protect their work in the AI age. Umair Kazi from the Authors Guild suggested that licensing can ensure legal training and provide rights holders control over how their works are used by AI.

Existing agreements, like those between Perplexity AI and media outlets, show that licensing models could work. However, due to ongoing litigation with major AI firms, licensing agreements are rare.

Kazi concludes that finding a balance through licensing could offer a viable solution moving forward.