The view is that reproduction of copyrighted works during the LLM learning stage constitutes fair use.
AI-generated summary
The U.S. Department of Justice has submitted a brief to the district court expressing the government's position in a copyright lawsuit involving OpenAI.
On September 1 (local time), the U.S. Department of Justice filed an opinion in the U.S. District Court for the Southern District of New York regarding multiple copyright infringement lawsuits involving U.S. OpenAI.
A written opinion is a document that expresses the government's view regarding the interests of the United States in pending litigation. Although it is not binding on the court, it may influence judicial decisions.
This includes lawsuits brought against OpenAI and other companies by multiple newspaper companies, publishers, and authors, including The New York Times.
The opinion states that reproduction of copyrighted works during the LLM study stage is a "transformative" use (not a substitute for the original work, but for a different purpose) and constitutes fair use.
When LLMs study texts, they explain that the purpose is to use them as material for learning statistical patterns such as vocabulary and syntax, and not to reproduce the expressions themselves of works. Learning and output should be considered separately, and even if a model can output part of a copyrighted work, it should be separated from the determination of whether learning is fair use.
He also cited the impact on national security. There is a danger that if AI learning requires permission from rights holders, AI development in the United States will stall, while it will give a competitive advantage to competing countries that are not subject to such restrictions.
He also took a stand against the interpretation of the law that requires paid licenses for learning. He points out that only the largest tech companies may be able to bear the licensing fees, which would hinder competition in the LLM market and skew the profits to legacy media with large amounts of content. It is argued that it is not in the public interest for large companies to have an oligopoly on learning.
The New York Times sued OpenAI and Microsoft in December 2023. The case is still pending, combined with lawsuits brought by other newspaper companies and book authors. In December 2025, the newspaper also filed a lawsuit against US AI search company Perplexity.

It has been revealed that an Osaka High Court clerk mistakenly added about 6,000 unrelated lawyers and court staff to a group on the online conferencing system ``Teams'' when sharing civil trial materials. The group was deleted before the materials were uploaded, and no information leaks have been confirmed.

The Supreme Court has decided to dismiss the former private school teacher's appeal in a lawsuit filed by a former private school teacher who was fired for indecent acts against students, seeking to have his teaching license revoked without a reinvestigation. The second instance judgment that the disposition was legal has become final.

In a lawsuit filed by a former postal worker who was fired for driving under the influence, the Fukuoka District Court ruled on the 3rd to invalidate the dismissal as an "abuse of disciplinary power" and order Japan Post to pay unpaid wages.

Regarding the Nagoya District Court's decision to deny bail to a defendant accused of stealing a stuffed animal, the Supreme Court's 2nd Petty Bench ruled that the bail was illegal as the defendant did not show any substantive reason for fearing that he would destroy the criminal evidence.

The first trial of five oil marketing companies, including Higashi Nippon Usami, who were accused of violating the Antimonopoly Act in a pricing cartel case involving the sale of diesel oil to corporations in Tokyo, was held on the 3rd at the Tokyo District Court, and all five companies admitted to the charges. Prosecutors pointed out that the price was agreed upon at regular meetings.

The National Police Agency's panel of experts has recommended the introduction of a system to remotely capture and analyze smartphone communications as a measure against anonymous and mobile criminal groups. We are calling for a process to obtain a court warrant, limited to serious crime prevention, with consideration given to the confidentiality of communications as stipulated in the Constitution.