argument: Notizie/News - Intellectual Property Law
Source: Legis1
Congress is weighing whether to introduce copyright rules for AI training or leave existing law to develop through litigation. A Congressional Research Service report examines permission to use copyrighted training material and protection for works involving AI. In June 2025, the Northern District of California found the training uses before the courts in Bartz v. Anthropic and Kadrey v. Meta Platforms to be fair use. Those conclusions concerned the particular uses examined and did not establish that all generative AI training is lawful. The decisions took different approaches to potential market harm, including competition between AI outputs and copyrighted works. In Bartz, the court also distinguished training with lawfully obtained books from Anthropic’s acquisition and retention of pirated copies.
The Trump administration maintains that training AI on copyrighted material does not infringe copyright, while acknowledging opposing legal arguments. The White House framework recommends allowing courts to resolve fair use and exploring collective licensing arrangements compatible with antitrust law. Five congressional measures identified in the coverage remain at the committee-referral stage. The proposals address additional civil remedies, access to information about training materials, and disclosures by AI developers. Separately, the D.C. Circuit held in Thaler v. Perlmutter that copyright requires human authorship, rejecting registration of a work attributed entirely to AI. The Copyright Office’s 2023 guidance likewise excludes independently generated AI material while allowing protection for sufficient original human contributions in works containing AI elements.