Noi perspective referitoare la antrenarea modelelor de inteligență artificială și dreptul de autor: jurisprudență, studii și propuneri tehnice, precum și rapoarte adoptate de autorități în UE și SUA
New perspectives on the training of artificial intelligence models and copyright: case law, studies and technical proposals, as well as reports adopted by authorities in the EU and the USA

Summary/Abstract: This article continues and develops the analysis published previously on the application of copyright exceptions and limitations in the training of generative artificial intelligence models. This article examines, from an EU-US comparative law perspective, the normative, judicial, and institutional developments during 2025-2026 regarding the use of copyrighted content in training generative artificial intelligence models. The analysis highlights the transition from debates on the applicability of text and data mining (TDM) exceptions and the fair use doctrine towards a systemic governance framework centered on transparency, licensing, and technical evidence. The study reviews official reports (EUIPO, US Copyright Office, European Parliament), recent case law (Thomson Reuters v. Ross Intelligence, Andersen v. Stability AI, Getty Images v. Stability AI, Kadrey v. Meta), opt-out mechanisms, the emerging licensing market, and technical self-protection tools.

• Details
• Contents