In the Southern District of New York, the author Julian Sancton filed a proposed copyright class action against OpenAI, its subsidiaries, and Microsoft.
The Complaint is below:
Using ChatGPT, we created a Table summarizing the 2 counts in the Complaint.
Sancton v. OpenAI: Counts, Specific Allegations, and Named Defendants
| Count | Specific Allegations | Named Defendants |
| Count 1: Copyright Infringement (Direct) | OpenAI and Microsoft created unlicensed reproductions of copyrighted works, including those owned by Julian Sancton and the putative class, during the training and fine-tuning of their GPT models. The training datasets for GPT-3, GPT-3.5, and GPT-4 included 45 terabytes of data from sources like Common Crawl, WebText2, Books1, Books2, and Wikipedia. This large-scale copying process involved reproducing copyrighted material, including nonfiction books, to train and fine-tune the AI models. Sancton’s book “Madhouse at the End of the Earth” was among the works used.. | OpenAI, Inc., OpenAI GP LLC, OpenAI, LLC, OpenAI Global LLC, OAI Corporation, LLC, OpenAI Holdings, LLC, Microsoft Corporation |
| Count 2: Contributory Infringement | Microsoft materially contributed and facilitated OpenAI’s direct infringement alleged in Count I by providing billions of dollars in investments and designing, creating, and maintaining the bespoke supercomputing system that OpenAI used to maintain and copy the copyrighted works owned by Plaintiff and the proposed Class. Millions of copyrighted works, including tens of thousands of nonfiction books, were copied and ingested for the purpose of “training” Defendants’ GPT model.. | Microsoft Corporation, OpenAI, Inc., OpenAI GP LLC, OpenAI, LLC, OpenAI OpCo LLC, OpenAI Global LLC, OAI Corporation, LLC, OpenAI Holdings, LLC |