JAKARTA - A US federal court has approved a settlement of a copyright lawsuit against Anthropic worth 1.5 billion US dollars. The AI company can now begin paying compensation to authors and publishers for about 500,000 works.
TechCrunch reported on Tuesday, July 21, citing Reuters, that the final approval was given by Judge Araceli Martinez-Olguin of the US District Court for the Northern District of California on Monday.
The payment is set at 3,000 US dollars for each work. The funds will be distributed to the authors and publishers who hold the rights to the books.
The settlement is believed to be the largest in US copyright law history. Despite its large value, many writers and creators have not considered it a victory.
The problem lies in another part of the judge's ruling. Judge William Alsup, who has now retired, stated that the use of copyrighted text to train an AI model could fall under the category of fair use.
Fair use is a limited exception to copyright law. Its application is assessed on the basis of the circumstances of each case.
Alsup's ruling is considered important for the AI industry because it touches on the legal basis for using books to train models. However, the judge distinguished the use of works from the way Anthropic obtained them.
Anthropic builds a collection of books to train AI from two sources. Some of the books are purchased and then scanned themselves. This way is considered not to violate the law.
Some of the other books were downloaded from the piracy sites Library Genesis and Pirate Library Mirror. Alsup said the method was still illegal and could be taken to trial.
Anthropic then chose to make peace to avoid trial and the amount of compensation that the jury might determine.
The final approval closes this lawsuit. However, the decision does not automatically resolve the copyright debate across the AI industry.
Alsup's ruling comes from a single district court. Because Anthropic chose a settlement, the case will not be reviewed by an appeals court and will not set a binding precedent.
Judges in other cases can still make different decisions based on the facts of each case. A number of similar lawsuits are still being faced by Google, Meta, Midjourney, and OpenAI.
Last week, a group of publishers and authors including Hachette, Cengage, Elsevier, Scott Turow, and S.C.R.I.B.E. also sued Google. They accused the company of using copyrighted works to train the Gemini AI platform.
The English, Chinese, Japanese, Arabic, and French versions are automatically generated by the AI. So there may still be inaccuracies in translating, please always see Indonesian as our main language. (system supported by DigitalSiber.id)
