In a historic development, AI company Anthropic has agreed to pay $1.5 billion USD to settle a class-action lawsuit launched by a group of authors that include renowned authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson.
The plaintiffs accused the company of illegally using pirated copies of their books to train its AI model, Claude, a common practice for AI programs of its kind. Court filings revealed that Anthropic maintained a central library of over seven million pirated books, obtained from sources like Library Genesis and the Pirate Library Mirror. It was ruled that while training with legally obtained materials might be considered “exceedingly transformative” and fall under fair use, the mass downloading and storage of pirated works was not. In response, Anthropic opted for the $1.5 billion agreement, which would yield approximately $3,000 per book for around 500,000 works and to destroy the pirated book files
Currently, Judge Alsup has paused approval of the settlement and another hearing is scheduled. If the settlement is approved, it would be the largest publicly reported copyright settlement in history, and a landmark development in AI related cases.
Implication and What Indonesia Must Prepare for
The Anthropic case is important development in an ever-growing crisis, which is AI companies’ reliance on copyrighted content that is often without permission or compensation. This problem raises a plethora of questions such as how can creators protect their work when machines are trained on data they didn’t authorize? Who owns the outputs of generative AI models trained on copyrighted material? Are existing copyright laws sufficiently equipped to handle the challenges posed by large-scale AI training? What responsibility do AI companies have in identifying and filtering out copyrighted content during training? Etc
While the potential settlement and the magnitude may push for more accountability that AI companies to be more responsible for how they source their training data, it is still an urgent matter for governments, even outside of the US,. to adapt its regulatory structures to better suit the development for AI.
Indonesia, a country with a large population rich in both traditional and modern cultural heritage, needs to always keep a keen eye on cases like this and all AI related developments. Antropic case is one of many cases worth studying.
Addressing the challenges of AI training on copyrighted content requires a more nuanced perspective at copyright laws to better fit the contemporary reality. Laws should push AI developers to be more transparent about the sources of their training data, especially when it involves local content and perhaps develop a licensing system for AI to process copyrighted material.
Furthermore, royalties is a complicated issue where a more innovative compensation model could be the future—such as collective licensing or blockchain-based royalty tracking that ensure creators receive tangible benefits. Such mechanisms may not be common place now, but is one of the many challenging possible developments that must be adapted to.
Alongside this, raising awareness among Indonesian authors, artists, academics, and institutions about how their work could be used and what rights they have is essential. As a strong regulatory system and a conscious public that compliments each other. Empowering creators with this knowledge will help them better navigate the evolving digital landscape.
For more information regarding this topic and other IP related inquiries, contact us at ambadar@ambadar.co.id. Our team of highly qualified professionals are more than ready to assist you.
Sources:
https://www.bbc.com/news/articles/c5y4jpg922qo
https://www.kompas.id/artikel/revolusi-ai-atau-perampokan-hak-cipta-pelajaran-dari-skandal-anthropic
