Anthropic's $1.5 Billion Copyright Settlement Approved Same Week It Faces New Neural Network Patent Infringement Suit from University of Tennessee

On July 20, 2026, the U.S. District Court for the Northern District of California officially approved a $1.5 billion class-action copyright settlement between artificial intelligence company Anthropic and a group of authors and publishers. The judgment, issued by Judge Araceli Martínez‑Olguín, marks the formal entry into effect of the largest monetary copyright class-action settlement in U.S. history.

The case originated from Anthropic's downloading of approximately 482,000 copyrighted books from pirate websites Library Genesis and Pirate Library Mirror to train its Claude large language model. In June 2025, the presiding judge, William Alsup, issued a split ruling, holding that training AI on lawfully purchased books constituted fair use, but that downloading books from pirate sites did not qualify as fair use. That ruling exposed Anthropic to statutory damages of up to $150,000 per work, with total potential exposure potentially reaching hundreds of billions of dollars, prompting the parties to reach a settlement agreement in September 2025.

Under the settlement agreement, Anthropic will pay a non‑refundable settlement fund of $1.5 billion. Rights holders of the approximately 482,000 works on the list are eligible to receive approximately $3,000 per work—an amount four times the statutory minimum of $750 for copyright infringement. Additionally, Anthropic must destroy all copies of the books downloaded from pirate sites, as well as any derivative copies. Class members release only claims related to Anthropic's past reproduction of their works, and do not release claims concerning AI outputs or any future conduct.

Although this case sets a record for settlement amounts in U.S. copyright litigation, because it is a settlement rather than a judgment, it does not establish binding legal precedent. Anthropic has emphasized that the court previously found training AI on books to be fair use, and that legal position remains unchanged by the settlement.

In the same week that the $1.5 billion copyright settlement was approved, Anthropic faced new legal challenges. The University of Tennessee Research Foundation (UTRF) filed suit on July 20, 2026, in the U.S. District Court for the District of Delaware, alleging that Anthropic had used two neural network patented technologies invented by a university professor without authorization. This is the first patent infringement lawsuit Anthropic has faced since its founding.

The two patents‑in‑suit are U.S. Patent Nos. 10,019,470 and 10,095,718. The ’470 patent, entitled “Method and Apparatus for Constructing, Using and Reusing Components and Structures of an Artificial Neural Network,” was issued in July 2018; the ’718 patent relates to methods for constructing Dynamic Adaptive Neural Network Arrays (DANNA) and was issued in October 2018. These technologies stem from work conducted since 2014 by the TENNLab research group at the University of Tennessee in the field of brain‑inspired computing, encompassing significant contributions to artificial intelligence, machine learning, neuromorphic computing, and neuroscience‑inspired computing.

The complaint specifically targets Anthropic's coding agent tool, Claude Code, and its underlying software architecture, alleging that Anthropic's products implement the neuromorphic network construction methods described in the patents, including a background execution scheduling system and a memory integration engine related to one of the patents. The Foundation stated in strong terms that Anthropic's "disregard for the intellectual property rights of others in the development of its products is not limited to its use of copyrighted materials."

An Anthropic spokesperson responded that the company disagrees with the allegations and plans to vigorously defend itself. The China Intellectual Property Lawyer Network will continue to follow further developments.