OpenAI Is Intertwined With Microsoft and GitHub

VII. FACTUAL ALLEGATIONS

K. OpenAI Is Intertwined with Microsoft and GitHub

167. OpenAI, Inc. is a nonprofit corporation founded in December 2015 by a group that included Greg Brockman, Ilya Sutskever, and other AI researchers; Elon Musk, CEO of Tesla; and Sam Altman, president of Y Combinator, a tech-startup incubator with hundreds of companies in its portfolio. Musk and Altman served as co-chairs of OpenAI, Inc. One of OpenAI, Inc.’s current board members is Reid Hoffman, founder of LinkedIn, which is now a Microsoft subsidiary. Mr. Hoffman is also a member of the Microsoft Board of Directors.

168. Less than a year later, in November 2016, OpenAI first partnered with Microsoft. It described the partnership as follows: “We’re working with Microsoft to start running most of our large-scale experiments on Azure. This will make Azure the primary cloud platform that OpenAI is using for deep learning and AI, and will let us conduct more research and share the results with the world.”

169. Initially, OpenAI, Inc. held itself out as a “non-profit artificial intelligence research company” that sought to shape AI “in the way that is most likely to benefit humanity as a whole.”

170. OpenAI, Inc. reportedly secured $1 billion in initial funding, from sources that were largely not disclosed, but included at least most of its founders.

171. OpenAI, Inc. obtained its initial source of training data from its founders’ companies. According to reporting at the time, Musk and Altman planned to “pool[] online data from their respective companies” to serve as training data for OpenAI, Inc. projects. Musk planned to contribute data from Tesla; Altman planned to have Y Combinator companies “share their data with OpenAI.” [29]

172. In February 2019, Altman created OpenAI, LP, a for-profit subsidiary of the nonprofit entity OpenAI, Inc. The new OpenAI, LP entity would serve as a vessel for accepting traditional outside investment in exchange for equity and distributing profits.

173. In July 2019, OpenAI, L.P. accepted a $1 billion investment from Microsoft. In addition to cash, Microsoft would become the exclusive licensor of certain OpenAI, LP products (including GPT-3, described below in Paragraph 176). Also, as part of this alliance, OpenAI, LP would use Microsoft’s cloud-computing platform, Azure, exclusively to develop and host its products. Some portion of Microsoft’s investment was paid in credits for use of Azure rather than cash. Finally, Microsoft and OpenAI agreed to “jointly build new Azure AI supercomputing technologies.”

174. Azure is a major growth area for Microsoft. In its most recent earnings report on October 25, 2022, “Azure and other cloud services” grew by 35% from the previous quarter, more than any other product.30 Azure has grown rapidly since Microsoft began its partnership with OpenAI in 2016. Its revenue grew by 50% or more every quarter from 2016 through the first three quarters of 2020.

175. In May 2020, Microsoft and OpenAI announced they had jointly built a supercomputer in Azure that would be used exclusively by OpenAI to train its AI models. Microsoft’s influence over and frequent collaboration with OpenAI has led some to describe Microsoft as “the unofficial owner of OpenAI.”[31]

176. One of OpenAI’s projects is GPT-3, a so-called “large language model” designed to emit naturalistic text. When researchers noticed that GPT-3 could also generate software code, they started studying whether they could make a new AI model specifically trained for this purpose. This project became known as Codex.

177. Sometime after July 2019, OpenAI and Microsoft began collaborating on a codecompletion product for GitHub that would use Codex as its underlying model. This product became known as Copilot.

178. On September 28, 2022, OpenAI released an image-generation AI called DALLE-2. Much like Copilot, DALL-E-2 removes any attribution and/or copyright notice from the images it uses to create derivative works. Like with Codex, here, OpenAI ignores the rights of the owners of copyrights to images it has ingested.

179. In another joint project, Microsoft and OpenAI recently launched a preview of a product called “Azure OpenAI Service.”[32] This service will “Leverage large-scale, generative AI models with deep understandings of language and code to enable new reasoning and comprehension capabilities for building cutting-edge applications. Apply these coding and language models to a variety of use cases, such as writing assistance, code generation, and reasoning over data. Detect and mitigate harmful use with built-in responsible AI and access enterprise-grade Azure security.”

About HackerNoon Legal PDF Series: We bring you the most important technical and insightful public domain court case filings.

This court case 4:22-cv-06823-JST retrieved on August 26, 2023, from Storage Courtlistener is part of the public domain. The court-created documents are works of the federal government, and under copyright law, are automatically placed in the public domain and may be shared without legal restriction.