Textbook of AI

OpenAI is the artificial-intelligence organisation founded in December 2015 by Sam Altman, Elon Musk, Ilya Sutskever, Greg Brockman, Wojciech Zaremba, John Schulman, Andrej Karpathy, Pieter Abbeel and others. It was founded as a non-profit research organisation aiming to develop safe artificial general intelligence; in 2019 it transitioned to a "capped-profit" structure under the governance of the original non-profit, allowing investment from Microsoft and others while preserving (in principle) the original safety mission.

OpenAI has shipped a long sequence of products and research results: GPT-1 (2018), GPT-2 (2019), GPT-3 (2020), DALL-E (2021), Codex (2021), InstructGPT (2022), ChatGPT (November 2022, the consumer product that brought LLMs to mass attention), GPT-4 (2023), DALL-E 3 (2023), Sora (text-to-video, 2024), GPT-4o (2024), o1 (reasoning model, 2024), o3 (2024).

The November 2023 OpenAI board crisis, the board's brief removal of Sam Altman as CEO, the threatened mass-resignation of nearly all OpenAI employees, and Altman's reinstatement within five days, was one of the most consequential corporate-governance episodes in AI history and the moment at which the organisation's tension between commercial pressure and safety mission became most public.

OpenAI partners closely with Microsoft, which has committed approximately USD 13 billion in funding and which deploys OpenAI's models through Azure and the Microsoft Copilot product line. The relationship has been mutually transformative: OpenAI gets the compute and distribution it needs; Microsoft gets a frontier-model lead over Google.