Vai al contenuto

Cookie di analisi

Usiamo Google Analytics e Google Tag Manager per contare le visite e capire quali pagine sono utili. Se accetti, Google Analytics può impostare cookie per riconoscere le tue visite successive. Se rifiuti, non imposta alcun cookie, ma Google riceve comunque un segnale senza cookie per ogni pagina che visiti. Non inviamo a Google il tuo nome, il tuo indirizzo e-mail né i dati della tua registrazione, e puoi cambiare la tua scelta in qualsiasi momento da «Impostazioni cookie» in fondo alla pagina. Privacy

Questa traduzione dell'interfaccia è una bozza e non è stata rivista da una persona madrelingua. L'inglese è la versione di riferimento. English

Microsoft told staff to keep OpenAI scraping project 'secret', unsealed filing shows

16 ott 2026 · Raccolto automaticamente da fonti pubbliche. Non revisionato da una persona.

Questo contenuto è disponibile solo nella lingua originale (inglese).

What happened

A partly unsealed version of news publishers' motion for summary judgment in the consolidated copyright case against Microsoft and OpenAI (In re OpenAI, Inc. Copyright Infringement Litig., S.D.N.Y. No. 1:25-md-03143) was released on October 16, 2026, disclosing new details about a web-scraping operation code-named "Project Mango." According to the publishers' filing, as reported by Bloomberg Tax, OpenAI paid Microsoft to develop and operate a crawler to copy webpage content that was used to train models including GPT-4o, and Microsoft ordered its employees to keep the project "secret."

What the filing states

  • The data allegedly included copies of more than 160,000 of the publishers' works. How much OpenAI paid Microsoft and the total number of webpages scraped remain under seal.
  • The publishers estimate ChatGPT fields 47 million prompts a day seeking news content, derived from an OpenAI expert witness's assertion that 1.88% of all ChatGPT messages in July 2025 sought news, applied to 2.5 billion daily prompts.
  • The filing also describes "horse trading" data exchanges between the companies, including OpenAI giving Microsoft its full set of GPT-3 training data.
  • Microsoft originally argued the content should remain redacted but withdrew some of its requests on October 9 after the publishers objected.

Status

These are the publishers' claims in their summary-judgment motion, not court findings. The parties filed cross motions for summary judgment in September 2026, which will determine what claims related to AI training practices head to trial. Bloomberg Industry Group, which operates Bloomberg Law, has moved to intervene in the case for the limited purpose of opposing sealing and redactions.

Fonti