# The end of OpenAI? Sam Altman's dismissal unleashes a storm in the AI world Author: Fernando Nieto Lobato Original publication: 2023-11-22 Spanish original: https://estrategiabyaleph.substack.com/p/estrategia-8-juego-de-tronos-en-openai English URL: https://elcontemplador.github.io/estrategia-english/essays/008/ Status: Published translation This is a translation of the original Spanish essay published on 22 November 2023. Its claims, examples and forecasts retain that historical context. English publication: 2026-09-29 *Historical context: this article records the rapidly changing situation as the newsletter closed on 22 November 2023. Its updates, uncertainties and possible outcomes are retained as they appeared then.* Until well into Friday, when what many people — ourselves included — consider the business story of the year broke, this newsletter's main topic was going to be something else: specifically, [the fourth European AI Alliance Assembly, held in Madrid last week](https://digital-strategy.ec.europa.eu/es/events/4th-european-ai-alliance-assembly-leading-trustworthy-ai-globally). We cover it in two news items, and recordings of its two days of sessions are still available [here](https://www.youtube.com/live/ls36o6PZZHg?si=DBayVHcXxEjccq1I) and [here](https://www.youtube.com/live/--b2BY-IeBw?si=XDFtpXT7iJ27qJu7). But late on Friday evening in Spain, **the entirely unexpected dismissal of Sam Altman was announced. Altman was CEO of OpenAI, the company with the world's most advanced AI models — at least among those publicly available. So began an AI “*Game of Thrones*” that is still unfolding as we write, although Sam Altman's return to OpenAI appears to be confirmed. Its outcome, and the events themselves, will undoubtedly have a decisive influence on the development of artificial intelligence over the coming years.** Before attempting a brief summary of the intense sequence of events surrounding the dismissal and the subsequent twists in OpenAI's crisis, it is important to understand a few basic details about the company behind, among other things, the GPT models, ChatGPT — which will mark its first year on the market in a few days — and DALL·E 3. ## What is OpenAI? Background to the events GPT-4 itself provides a very interesting brief summary: > OpenAI is an artificial intelligence research organisation founded in 2015. **Its main focus is developing advanced and ethical AI technologies, with the aim of ensuring that AI's benefits are widely shared across society.** OpenAI is known for significant advances in deep learning and natural language processing. It seeks a balance between open collaboration and safety, advocating careful and ethical regulation of AI. The organisation has evolved from a non-profit model to a hybrid structure to sustain its research and development. *Translation note: this historical GPT-4 response is translated from the Spanish article; it is not an independently verified description or a quotation checked against an English original.* **OpenAI's origins as a research organisation with an entirely non-profit model are crucial to understanding the events of the past few days.** [OpenAI's website provides a diagram](https://openai.com/our-structure?stream=top) **of the organisation's extremely complex current structure**: its controlling board of directors, non-profit entities, holding company and capped-profit company, in which Microsoft holds 49%… [![OpenAI's November 2023 organisational diagram, showing board control, the non-profit, holding company, capped-profit company and Microsoft's minority ownership.](https://elcontemplador.github.io/estrategia-english/assets/images/05dc73b4-a7d2-4468-8c80-6fc52420bfc2_796x547.png)](https://substackcdn.com/image/fetch/$s_!AGSD!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F05dc73b4-a7d2-4468-8c80-6fc52420bfc2_796x547.png) *Diagram text and relationships: the board controls OpenAI, Inc., a 501(c)(3) public charity labelled “OpenAI Nonprofit”. This entity wholly owns and controls OpenAI GP LLC and owns a holding company together with employees and other investors. OpenAI GP LLC controls the holding company and OpenAI Global, LLC. The holding company is the majority owner of OpenAI Global, LLC, labelled “capped profit company”; Microsoft is its minority owner. The original diagram is already in English.* This complex and unusual structure underlies the events of these days. But **the external board overseeing OpenAI, which dismissed Sam Altman, has also played a decisive role**. Last week, its members included: - **Ilya Sutskever**: OpenAI co-founder and chief scientist, recognised for his work in deep learning and natural language processing. - **Adam D'Angelo**: CEO of Quora and formerly CTO of Facebook. - **Tasha McCauley**: technology entrepreneur, co-founder and CEO of Fellow Robots, a robotics company serving the retail sector. - **Helen Toner**: director of strategy at Georgetown's Center for Security and Emerging Technology, and an expert in AI policy and national security. Alongside them were the dismissed Sam Altman and Greg Brockman, who was removed from his role on the board he chaired — and apparently was not even informed beforehand of Altman's dismissal — before also resigning as the company's president. ## A chronology of the dismissal and return In very brief terms — [you can read a more detailed chronology at Xataka, for example](https://www.xataka.com/empresas-y-economia/cronologia-crisis-openai-todo-que-ha-pasado-despido-sam-altman-hoy) — events unfolded as follows: **Friday, 17 November**: [OpenAI announces the dismissal of CEO Sam Altman](https://openai.com/blog/openai-announces-leadership-transition). The board cites a lack of transparency in his communication as the reason. Mira Murati is appointed interim CEO. OpenAI president Greg Brockman resigns in solidarity, as do three senior researchers. **Saturday, 18 November**: the news surprises major investors, including Microsoft, which says it learned of the dismissal only one minute before it became public. OpenAI chief operating officer Brad Lightcap rules out misconduct as the reason for the dismissal. The possibility of reinstating Altman and Brockman emerges. There are rumours that both are exploring a new AI project. **Sunday, 19 November**: negotiations over Altman's return fail, despite his visiting the company and taking a photograph with his visitor pass. Reports emerge that Emmett Shear, the former Twitch CEO, will replace Murati as interim CEO. OpenAI's board members reaffirm their decision to dismiss Altman. Before the stock market reopens on Monday, Microsoft announces that it has hired Altman and Brockman to lead a new advanced AI research team. **Monday, 20 November**: the possibility of a partnership between Microsoft and OpenAI is floated, which could stabilise the situation. Several OpenAI employees express support for Altman on social media through heart emojis. Ilya Sutskever, a board member and critic of Altman, expresses regret over the dismissal and a desire to reunify the company. OpenAI employees threaten to resign and join Altman at Microsoft unless the board reverses its decision and steps down. At the time of writing, more than 740 of the company's 770 employees have signed the letter against the board. Altman reaffirms his commitment to OpenAI's mission and to continuing its partnership with Microsoft. **Tuesday, 21 November**: late in the day in California, as we are closing this newsletter on Wednesday morning in Spain, an agreement in principle is reached for Sam Altman's return: [![OpenAI announces an agreement in principle for Sam Altman to return as CEO, with an initial board of Bret Taylor, Larry Summers and Adam D'Angelo.](https://elcontemplador.github.io/estrategia-english/assets/images/1ea34a83-adfa-490d-85df-1c6d59c2a364_886x312.png)](https://substackcdn.com/image/fetch/$s_!Sjms!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F1ea34a83-adfa-490d-85df-1c6d59c2a364_886x312.png) *OpenAI's announcement, transcribed from the original English screenshot:* > We have reached an agreement in principle for Sam Altman to return to OpenAI as CEO with a new initial board of Bret Taylor (Chair), Larry Summers, and Adam D'Angelo. > > We are collaborating to figure out the details. Thank you so much for your patience through this. Under this agreement, Bret Taylor, formerly co-CEO of Salesforce, and Larry Summers, US Treasury Secretary during Bill Clinton's presidency, join the new initial board. These are the facts of a crisis at what Musk has called the start-up with the greatest potential in history: with fewer than 800 employees, it was already valued at close to US$90 billion. Microsoft's “move” has not yet been completed. It could allow the company to acquire virtually all OpenAI's talent and its models at no cost and without regulatory interference. [Simply announcing the hiring of Altman and Brockman](https://twitter.com/satyanadella/status/1726509045803336122) sent Microsoft's shares to an all-time high, adding more than US$115 billion in value in the first hour of trading on Monday. It also prompted Elon Musk to tweet a subtitled video of “El Risitas”, which we include in our meme of the week section, capturing this extraordinary situation. **With the latest news and Sam Altman's return to OpenAI, it appears that Microsoft will not absorb the entire OpenAI team directly, but will very probably secure closer integration and a greater say in decisions.** How this ultimately ends will undoubtedly shape the coming years of AI development. Until now, Microsoft has had access to OpenAI's talent and technological advances, but in a fairly limited way. For example, a clause states that if AGI — artificial general intelligence, which experts increasingly see as approaching — is achieved, those OpenAI models would not pass directly to Microsoft but would be for the public good. That arrangement could now break down if virtually all OpenAI employees move directly to Microsoft, or could change if Sam Altman ultimately returns to lead OpenAI with Nadella's support. Competitors such as [Salesforce](https://twitter.com/Benioff/status/1726695914105090498) and [Nvidia](https://twitter.com/DrJimFan/status/1726490491452379250?t=pErEXxzJWRXYRubsCzq5yA&s=35) have publicly made substantial multimillion-dollar offers, even on X, formerly Twitter, to try to lure away talent seeking to leave OpenAI amid the upheaval. But **behind this Hollywood-style drama of the past few days, the various rumours and leaked internal accounts — the only information currently available about the deeper reasons — appear to point to two underlying issues. Both, particularly the second, are highly relevant to political approaches to AI and its regulation.** ## Non-profit purpose versus commercial deployment: research versus business One possible major disagreement, [as The Algorithmic Bridge discusses, for example](https://thealgorithmicbridge.substack.com/p/this-subtle-but-fundamental-conflict?r=2tyybn), appears to concern the imbalance that Altman was supposedly creating within a company whose foundations are non-profit. More and more resources were being devoted to commercialising ChatGPT and raising round after round of funding to launch new products, including the announcement at the developers' event in early November of a kind of app store for ChatGPTs. Altman's approach, which seemed aimed at placing OpenAI among the world's four or five largest technology companies, could clash sharply with the position defended by chief scientist Ilya Sutskever. He sought to prioritise the company's resources for research into new large models and, especially in recent months, the superalignment of AI systems to ensure their long-term safety. ## Those who want to move faster and those who want to slow down AI development and deployment Connected to the previous issue, **several analysts** — [Tomás Pueyo explores it at length in his newsletter, for example](https://unchartedterritories.tomaspueyo.com/p/openai-and-the-biggest-threat-in?utm_campaign=post&utm_medium=web) — **point to a degree of conflict, reinforced by these tensions, despite all OpenAI employees' commitment to safe AI. On one side are those seeking a slower, safer deployment of new AI foundation models**. [On Tuesday, The New York Times reported a dispute between Sam Altman and Helen Toner over an academic article by Toner that raised doubts about OpenAI's safety work](https://www.nytimes.com/2023/11/21/technology/openai-altman-board-fight.html). This could also help explain the appointment of Emmett Shear as interim CEO. Although he describes himself as a techno-optimist, he has said in recent interviews that AI's pace should drop from 10 to 1 or 2, **because the risks of AGI are too great to get it wrong. On the other side are those who believe new AI models must be developed safely but also very rapidly: the capabilities they would give humanity to address many of its greatest challenges — climate change, curing diseases, energy — would far outweigh the possible risks, and every additional day before these advances are achieved is another day of suffering for humanity.** Ilya Sutskever may favour a slower, safer deployment of models, in contrast to Sam Altman's commercial appetite. This could stem from his work in OpenAI's laboratory, where several tweets by him and Altman in recent months point to significant, even unexpected, advances in new models' capabilities, and from [his work on AI superalignment](https://www.youtube.com/watch?v=Ft0gTO2K85A): the aim is for future AGIs to feel positive emotions towards humans, rather than being mere “intelligences”. This also connects with the previous issue. It remains to be seen how all this will develop, depending on whether OpenAI retains control of its models — impossible if 95% of its employees leave — or whether the situation is reversed, as now appears likely with Altman's return. The possible de facto acquisition of OpenAI by Microsoft — which now seems unlikely to happen, at least in that form — would be a business masterstroke, but also raises many questions about the world's most powerful AI being monopolised by its largest companies. Microsoft had already announced annual investment of more than US$50 billion in AI server infrastructure from 2024. With OpenAI, Microsoft would now clearly move ahead of Google, which previously absorbed DeepMind, then the leading AI company, to turn it into its AI laboratory. These corporations may be slower and more careful in deploying large AI models, although there is fierce competition between them and with perhaps slightly smaller but still well-resourced players such as Amazon, Tesla, Salesforce and Nvidia. Yet their size and capabilities also make them quite capable of establishing virtual monopolies, as we have already seen with computer and mobile operating systems, business applications, browsers and app stores. **From a political perspective, the moves of recent days should prompt close attention to the underlying issues and to how political action can respond effectively to the challenges posed by the very structure of AI development and deployment, while continuing to encourage the rapid, responsible development of a technology that, used well, could usher in a golden age for humanity.** Fernando Nieto Lobato *Director of Digital Innovation, Institución Educativa ALEPH*