Zum Inhalt

Diese Oberflächenübersetzung ist ein Entwurf und wurde nicht muttersprachlich geprüft. Englisch ist die Quellfassung. English

Unsealed publishers' brief in consolidated NYT copyright MDL details internal Microsoft and OpenAI documents on five stages of copying

17.09.2026 · Automatisch aus öffentlichen Quellen erfasst. Nicht von einer Person geprüft.

Dieser Beitrag liegt nur in seiner Originalsprache vor (Englisch).

What happened: After all three sides moved for summary judgment on 2026-09-04 in the consolidated AI copyright litigation (MDL 25-md-3143, S.D.N.Y., Judge Sidney H. Stein), the news-publisher plaintiffs (The New York Times, Daily News titles, Ziff Davis, the Center for Investigative Reporting and The Intercept) filed the public, less-redacted version of their 92-page combined summary judgment brief on 2026-09-17. The brief asks the court to rule that OpenAI and Microsoft infringed copyright at five stages — acquisition, training, grounding, outputs and inter-company 'horse trading' (Project Taxi/Project Mango data exchanges) — and that fair use excuses none of them. It quotes internal documents extensively: a Microsoft director of applied science called AI ingestion of creators' work 'an astonishing theft of unprecedented proportions' and possibly 'the largest theft of labor in human history'; OpenAI's head of ChatGPT called the company's products 'largely substitutive, period'; Microsoft CEO Satya Nadella testified that paywalled content should be licensed by anyone using it for training or grounding. The brief cites internal Microsoft data showing Copilot cut publisher click-through rates by up to 94% versus Bing search, alleges OpenAI sidestepped robots.txt and paywalls, and says OpenAI abandoned its promised Media Manager opt-out tool in 2024.

Why it matters to creators: This is the fullest public accounting yet of what news publishers say AI companies knew about their own copying — including private admissions that outputs substitute for the original journalism. If the court accepts the five-stage infringement theory with per-article statutory damages, licensing publisher content becomes the industry norm rather than the exception; if fair use prevails, text creators will likely need legislation, not litigation, for compensation.

Not yet known: Everything quoted comes from the plaintiffs' brief, not a judge's findings — OpenAI and Microsoft argue fair use and contest the framing. No ruling on the summary judgment motions has issued. Grounding/output claims are pressed against Microsoft only, with the corresponding claims against OpenAI deferred pending a sanctions motion.

Quellen